Understanding of Harmonics in Power System Mack Grady

Grady Chapter 1.
Introduction Page 1-1

June 2006
Understanding Power System Harmonics

Prof. Mack Grady
Dept. of Electrical & Computer Engineering
University of Texas at Austin
grady@mail.utexas.edu, www.ece.utexas.edu/~grady
Table of Contents
x Chapter 1. Introduction
x Chapter 2. Fourier Series
x Chapter 3. Definitions
x Chapter 4. Sources
x Chapter 5. Effects and Symptoms Chapter 8a. System Matrices and
x Chapter 6. Conducting an Investigation Simulation Procedures
x Chapter 7. Standards and Solutions
x Chapter 8. Case Studies Chapter 9. Homework Problems
Chapter 10. Acknowledgements
x Appendix A1. Fourier Spectra Data
x Appendix A2. PCFLO User Manual and Files
x Appendix A3. Harmonics Analysis for Ships and Industrial Power Systems (HASIP)
User Manual and Files
We are interested mostly in harmonics 1 through the 25th. (25 * 60 =
1. Introduction 1500Hz). But most harmonic problems are due to the 5th and 7th, and also
the 3rd. Modeling accuracy is not good beyond the 25th.
Power systems are designed to operate at frequencies of 50 or 60Hz. However, certain types of
loads produce currents and voltages with frequencies that are integer multiples of the 50 or 60 Hz
fundamental frequency. These higher frequencies are a form of electrical pollution known as
power system harmonics.
Power system harmonics are not a new phenomenon. In fact, a text published by Steinmetz in
1916 devotes considerable attention to the study
y of harmonics in three-phase power systems. In
Steinmetz’s day, the main concern was third harmonic currents caused by saturated iron in
transformers and machines. He was the first to propose delta connections for blocking third
harmonic currents.
After Steinmetz’s important discovery, and as improvements were made in transformer and
machine design, the harmonics problem was largely solved until the 1930s and 40s. Then, with
the advent of rural electrification and telephones, power and telephone circuits were placed on
common rights-of-way. Transformers and rectifiers in ppower systems
y p
produced harmonic
currents that inductively coupled into adjacent open-wire telephone circuits and produced
audible telephone interference. These problems were gradually alleviated by filtering and by
minimizing transformer core magnetizing currents. Isolated telephone interference problems still
occur, but these problems are infrequent because open-wire telephone circuits have been
replaced with twisted pair, buried cables, and fiber optics.
Page 1-1
Grady Chapter 1. Introduction Page 1-2
June 2006
Today,y, the most common sources of harmonics are power electronic loads such as adjustable-
speed drives (ASDs) and switching power supplies. Electronic loads use diodes, silicon-
controlled rectifiers (SCRs), power transistors, and other electronic switches to either chopp
p
waveforms to control power, , or to convert 50/60Hz AC to DC. In the case of ASDs,, DC is then
converted to variable-frequency
f AC to control motor speed. Example uses of ASDs include
chillers and pumps.
A single-phase power electronic load that you are familiar with is the single-phase light dimmer
shown in Figure 1.1. By adjusting the potentiometer, the current and power to the light bulb are
controlled, as shown in Figures 1.2 and 1.3.
Power electronic devices are
Light used as on and off switches
a bulb b
Triac
(front view)
3.3kȍ
MT2
+ 250kȍ c G Triac
120Vrms AC linear MT1
– pot
Bilateral trigger
0.1μF diode (diac) MT1 MT2 G
n
Light Light
a bulb b a bulb b
+ + 0V – + + + Van – +
Van Van Van 0V
– n – – n –
Before firing, the triac is an open switch, After firing, the triac is a closed
so that practically no voltage is applied switch, so that practically all of Van
across the light bulb. The small current is applied across the light bulb.
through the 3.3k resistor is ignored in
this diagram.
Figure 1.1. Triac light dimmer circuit
Page 1-2
June 2006
Į = 30º Į = 90º
Current
Current
0 30 60 90 120 150 180 210 240 270 300 330 360 0 30 60 90 120 150 180 210 240 270 300 330 360
Angle Angle
Į = 150º
Current
0 30 60 90 120 150 180 210 240 270 300 330 360
Angle
Figure 1.2. Light dimmer current waveforms for firing angles

Į = 30º, 90º, and 150º
Efficient control of power
1
0.9
0.8
0.7
Half the wave,
0.6 half the power
0.5
P
0.4
0.3
0.2
0.1
0
0 30 60 90 120 150 180
Alpha
Figure 1.3. Normalized power delivered to light bulb versus Į
Page 1-3
June 2006
g dimmer is a simple
The light p example,
p but it represents
p two major benefits
f of power electronic
loads í controllabilityy and efficiency. The “tradeoff” is that power electronic loads draw
nonsinusoidal currents from AC power systems, and these currents react with system impedances
to create voltage harmonics and, in some cases, resonance. Studies show that harmonic
distortion levels in distribution feeders are rising as power electronic loads continue to proliferate
and as shunt capacitors are employed in greater numbers to improve power factor closer to unity.
Unlike transient events such as lightning that last for a few microseconds, or voltage sags that
last from a few milliseconds to several cycles, harmonics are steady-state, periodic phenomena
that produce continuous distortion of voltage and current waveforms. These periodic
nonsinusoidal waveforms are described in terms of their harmonics, whose magnitudes and
phase angles are computed using Fourier analysis.
Fourier analysis permits a periodic distorted waveform to be decomposed into a series containing
dc, fundamental frequency (e.g. 60Hz), second harmonic (e.g. 120Hz), third harmonic (e.g.
180Hz), and so on. The individual harmonics add to reproduce the original waveform. The
g
highest harmonic of interest in power systems is usually the 25th (1500Hz), which is in the low
audible range. Because of their relatively low frequencies, harmonics should not be confused
with radio-frequency interference (RFI) or electromagnetic interference (EMI).
Ordinarily, the DC term is not present in power systems because most loads do not produce DC
and because transformers block the flow of DC. Even-ordered harmonics are generally much
smaller than odd-ordered harmonics because most electronic loads have the property of half-
wave symmetry, and half-wave symmetric waveforms have no even-ordered harmonics.
The current drawn by electronic loads can be made distortion-free (i.e., perfectly sinusoidal), but
the cost of doing this is significant and is the subject of ongoing debate between equipment
manufacturers and electric utility companies in standard-making activities. Two main concerns
are
1. What are the acceptable levels of current distortion?
2. Should harmonics be controlled at the source, or within the power system?
Page 1-4
Grady Chapter 2. Fourier Series Page 2-1
June 2006
2. Fourier Series
2.1. General Discussion
Any physically realizable periodic waveform can be decomposed into a Fourier series of DC,
fundamental frequency, and harmonic terms. In sine form, the Fourier series is
f
i (t
(t ) I avg ¦ I k sin((kZ1t T k ) , (2.1)
k 1
and if converted to cosine form, 2.1 becomes
f
ii(((tt ) I avg os((kZ1t T k 90 o ) .
¦ I k cos(
co
k 1
I avg is the average (often referred to as the “DC” value I dc )). I k aare peak magnitudes of the
individual harmonics, Z o is the fundamental frequency (in radians per second), and T k are the
harmonic phase angles. The time period of the waveform is
2S 2S 1 For 60Hz, T = 16.67msec

T .
Z1 2Sf1 f1
The formulas for computing I dc , I k , T k are well known and can be found in any undergraduate
electrical engineering textbook on circuit analysis. These are described in Section 2.2.
Figure 2.1 shows a desktop computer (i.e., PC) current waveform. The corresponding spectrum
is given in the Appendix. The figure illustrates how the actual waveform can be approximated
by summing only the fundamental, 3rd, and 5th harmonic components. If higher-order terms are
included (i.e., 7th, 9th, 11th, and so on),
), then the PC current waveform will be perfectly
p y
reconstructed. A truncated Fourier series is actually a least-squared error curve fit. As higher
frequency terms are added, the error is reduced.
Fortunately,
y, a special
p pproperty
p y known as half-wave symmetry exists for most power electronic
loads. Have-wave symmetry exists when the positive and negative halves of a waveform are
identical but opposite, i.e.,
T
i ((tt ) i ((tt r ),
2
where T is the period. Waveforms with half-wave symmetry have no even-ordered harmonics.
It is obvious that the television current waveform is half-wave symmetric.
Page 2-1
June 2006
Amperes 0
-5
Sum of 1st, 3rd, 5th
5 3
Amperes
-5
Figure 2.1. PC Current Waveform,, and its 1st, 3rd, and 5th Harmonic Components
Note – in this waveform,, the harmonics are p
(Tote peakingg at the same time as the
fundamental. Most waveforms do not have this property. In fact, in many cases (e.g.
a square wave), the peak of the fundamental component is actually greater than the
peak of the composite wave.)
Page 2-2
June 2006
2.2 Fourier Coefficients
If function i (t ) is periodic with an identifiable period T (i.e., i (t ) i (t r NT ) ), then i (t ) can be

written in rectangular form as
f
2S
i (t ) I avg ¦ >ak cos(kZ1t ) bk sin(kZ1t )@ , Z1 T
, (2.2)
k 1
where
1 t o T
T
I avg
a ³
T to
i (t )d
dtt ,
2 to T
i (t ) cosskZ1t dt ,
T ³t o
ak
2 t o T
T
bk ³ i (t ) sin kZ1t dtt .
T to
The sine and cosine terms in (2.2) can be converted to the convenient polar form of (2.1) by
using trigonometry as follows:
a k cos(kZ1t ) bk sin(kZ1t )
a cos(kZ1t ) bk sin( kZ1t )

a k2 bk2 x k
a k2 bk2
ª º
« ak bk
a k2 bk2 x cos(kZ1t ) sin(kZ1t )»
« 2 2 2 2 »
«¬ a k bk a k bk »¼
a k2 bk2 x >sin(T k ) cos(kZ1t ) cos(T k ) sin( kZ1t )@ , (2.3)
where
ak bk
sin(T k ) , cos(T k ) .
2 2
a k bk a k2 bk2
ak
Applying trigonometric identity șk
sin( A B) sin( A) cos( B) cos( A) sin( B)

bk
Page 2-3
June 2006
yields polar form
a k2 bk2 x ssin(( kZ1t T k ) , (2.4)
where
sin((T k ) ak
tan((T k ) . (2.5)
cos((T k ) bk
2.3 Phase Shift
There are two types of phase shifts pertinent to harmonics. The first is a shift in time, e.g. the
±T/3 among balanced a-b-c currents. If the PC waveform in Figure 2.2 is delayed by 'T
seconds, the modified current is
5
delayed
Amperes
-5
Figure 2.2. PC Current Waveform Delayed in Time
f f
i (t 'T ) ¦ I k sin(kZ1 t 'T T k ) = ¦ I k sin(kZ1t kZ1'T T k )
k 1 k 1
f f
¦ I k sin kZ1t T k kZ1'T = ¦ I k sin kZ1t T k kT1 , (2.6)
k 1 k 1
where T1 is the phase lag of the fundamental current corresponding to 'T . The last term in
(2.6) shows that the individual harmonic phase angles are delayed by kT1 of their own degrees.
The second type of phase shift is in phase angle, which occurs in wye-delta transformers. Wye-
delta transformers shift voltages and currents by r 30 o , depending on phase sequence. ANSI
standards require that, regardless of which side is delta or wye, the a-b-c phases must be marked
so that the high-voltage side voltages and currents lead those on the low-voltage side by 30 o for
Page 2-4
June 2006
positive-sequence (and thus lag by 30 o for negative sequence). Zero sequences are blocked by
the three-wire connection so that their phase shift is not meaningful.
2.4 Symmetry Simplifications
Waveform symmetry
y y greatly
g y simplifies
p the integration
g effort required
q to developp Fourier
coefficients. Symmetry
y y arguments
g should be applied to the waveform after the average (i.e.,
DC) value has been removed. The most important cases are
x Odd Symmetry, i.e., i (t ) i (t ) ,
then the corresponding Fourier series has no cosine terms,
ak 0,
and bk can be found by integrating over the first half-period and doubling the results,
4 T /2
i (t ) sin kZ1t dt .
T ³0
bk
x Even Symmetry, i.e., i (t ) i (t ) ,
then the corresponding Fourier series has no sine terms,
bk 0,
and a k can be found by integrating over the first half-period and doubling the results,
4 T /2
i (t ) coskZ1t dt .
T ³0
ak
Important note – even and odd symmetry can sometimes be obtained by time-shifting the
waveform. In this case, solve for the Fourier coefficients of the time-shifted waveform, and
then phase-shift the Fourier coefficient angles according to (A.6).
T
x Half-Wave Symmetry, i.e., i (t r ) i (t ) ,
2
then the corresponding Fourier series has no even harmonics, and a k and bk can be
found by integrating over any half-period and doubling the results,
4 to T / 2
i (t ) coskZ1t dt ,
T ³t o
ak k odd,
Page 2-5
June 2006
4 to T / 2
i (t ) sin kZ1t dt ,
T ³t o
bk k odd.
Half-wave symmetry is common in power systems.
2.5 Examples
T/2
x Square Wave
V
By inspection, the average value is zero, and
the waveform has both odd symmetry and
half-wave symmetry. Thus, a k 0 , and
–V
4 to T / 2 T
v(t ) sin kZ1t dt , k odd.
T ³t o
bk
Solving for bk ,
4 T /2 4V t T /2 4V § § kZ1T · ·
V sin kZ1t dt coskZ1t t 0 ¨¨ cos¨ ¸ cos(0) ¸¸ .
T ³0
bk
kZ oT kZ1T © © 2 ¹ ¹
2S
Since Z1 , then
T
4V 2V
bk coskS 1 1 coskS , yielding
2kS kS
4V
bk , k odd.
kS
The Fourier series is then
f
4V 1 4V ª 1 1 º
v(t ) ¦ sin kZ1t «sin 1Z1t sin 3Z1t sin 5Z1t » . (2.7)
S k 1, k odd k S ¬ 3 5 ¼
1
Note that the harmonic magnitudes decrease according to .
k
Page 2-6
June 2006
x Triangle Wave T/2

V
By inspection, the average value is zero, and
the waveform has both even symmetry and
half-wave symmetry. Thus, bk 0 , and
–V
4 to T / 2 T
v(t ) coskZ1t dt , k odd.
T ³t o
ak
Solving for a k ,
4 T / 2 § 4t · 4V T / 2 16V T / 2
V ¨1 ¸ coskZ1t dt coskZ1t dt ³ t coskZ1t dt
T ³0 ³
ak
© T¹ T 0 T2 0
t T /2
4V § § kZ1T · · 16V t sin kZ1t 16V T / 2 sin kZ1t
¨¨ sin ¨ ¸ sin(0) ¸¸ 2 ³ dt
kZ1T © © 2 ¹ ¹ T kZ1 t 0 T2 0 kZ1
2V 4V 4V
sin kS sin kS 1 coskS , k odd.
kS kS k 2S 2
Continuing,
8V
ak , k odd.
k 2S 2
The Fourier series is then
f
8V 1
v(t ) ¦ cos kZ1t
2
S k 1, k odd k2
8V ª 1 1 º
cos 1Z1t cos3Z1t cos5Z1t » , (2.8)
2 «¬ 9 25 ¼
S
1
where it is seen that the harmonic magnitudes decrease according to .
k2
To convert to a sine series, recall that cos(T ) sin(T 90 o ) , so that the series becomes
Page 2-7
June 2006
v(t )
8V ª
S 2 «¬
1
9
1
25
º
sin 1Z1t 90 o sin 3Z1t 90 o sin 5Z1t 90 o » .
¼
(2.9)
T
To time delay the waveform by (i.e., move to the right by 90 o of fundamental),
4

subtract k x 90 o from each harmonic angle. Then, (2.9) becomes
v(t )
8V ª
S
«
2 ¬
1
·
sin 1Z1t 90 o 1 x 90 o sin 3Z1t 90 o 3 x 90 o ¸
9 ¹

1

sin 5Z1t 90 o 5 x 90 o » ,
25
º
¼

or
8V ª 1 1 1 º
v(t )
2 «¬
sin 1Z1t sin 3Z1t sin 5Z1t sin 7Z1t » . (2.10)
S 9 25 49 ¼
x Half-Wave Rectified Cosine Wave

T
I
The waveform has an average value and even
symmetry. Thus, bk 0 , and
4 T /2 T/2
i (t ) coskZ o t dt , k odd.
T ³0
ak
T/2
Solving for the average value, Not half-wave symmetric
t T /4
1 to T 1 T /4 I
I cosZ1t dt sin Z1t
T ³t o T ³ T / 4
I avg i (t )dt
Z oT t T / 4
I § Z1T Z1T · I ZT 1 S
¨ sin sin ¸ sin 1 sin .
2S © 4 4 ¹ S 4 S 2
I
I avg . (2.11)
S
Solving for a k ,
T /4
4 T /4 2I
I cosZ1t coskZ1t dt ³ cos1 k Z1t cos1 k Z1t dt
T ³0
ak
T
0
Page 2-8
June 2006
t T /4
2 I § sin 1 k Z1t sin 1 k Z1t ·
¨ ¸ .
T ¨© 1 k Z1 1 k Z1 ¸¹ t 0
For k 1 , taking the limits of the above expression when needed yields
§ § T ··
¨ sin ¨ 1 k Z1 ¸ ¸
2I 4 ¹ ¸ I sin S
a1 x lim ¨ ©
T (1 k )Z1 o 0¨ 1 k Z1 ¸ 2S
¨ ¸
© ¹
2I sin 1 k Z1 x 0 I sin 0
x lim . (2.12)
T (1 k )Z1 o 0 1 k Z1 2S
2I T I
a1 000 .
T 4 2
For k ! 1 ,
§ S S ·
sin 1 k sin 1 k
I ¨¨ 2 2
¸
¸.
ak (2.13)
S ¨ 1 k 1 k ¸
¨ ¸
© ¹
All odd k terms in (2.13) are zero. For the even terms, it is helpful to find a common
denominator and write (2.13) as
§ S S ·
¨
I¨
1 k sin 1 k 1 k sin 1 k ¸
ak 2 2 ¸ , k ! 1 , k even.
S¨ 2 ¸
¨ 1 k ¸
© ¹
Evaluating the above equation shows an alternating sign pattern that can be expressed as
f k 2
2I 1
ak
S
¦ 1 2 2
, k ! 1 , k even.
k 2, 4,6, k 1
The final expression becomes 1/(squared k) variation
f
II 2I
i (t ) cosZ1t ¦ 1k / 2 1 21 coskZ1t
S 2 S k 2,4,6, k 11
Page 2-9
June 2006
I I 2I ª1 1 1 º
cosZ1t «¬ 3 cos2Z1t 15 cos4Z1t 35 cos6Z1t »¼ . (2.14)
S 2 S
x Light Dimmer Current
The Fourier coefficients of the current waveform shown in Figure 1.2 can be shown to be
the following:
For the fundamental,
Ip ª D 1 º
a1 sin 2 D , b1 I p «1 sin 2D » , (2.15)
S ¬ S 2S ¼
where firing angle Į is in radians, and I p is the peak value of the fundamental current
when D 0q .
For harmonic multiples above the fundamental (i.e., k = 3, 5, 7 , … ),
Ip ª 1
ak « cos(1 k )D cos(1 k )S 1 cos(1 k )D cos(1 k )S º» , (2.16)
S ¬1 k 1 k ¼
Ip ª 1
bk « sin(1 k )S sin(1 k )D 1 sin(1 k )D sin(1 k )S º» . (2.17)
S ¬1 k 1 k ¼
The waveform has zero average, and it has no even harmonics because of half-wave
symmetry.
The magnitude of any harmonic k , including k = 1, is I k a k2 bk2 . Performing the

S
calculations for the special case of Į = radians (i.e., 90°) yields
2
Ip S2 Ip
I1 = 1 = 0.593 I p , and I 3 = = 0.318 I p ,
S 4 S
I3 1
= = 0.537.
I1 S2
1
4
The above case is illustrated in the following Excel spreadsheet.
Page 2-10
June 2006
Light_Dimmer_Fourier_Waveform.xls
Note – in the highlighted cells, the magnitude of I1 is computed to be 0.593 times the peak value of the fundamental current for the
D 0q case. The ratio of I3 to I1 is computed to be 0.537.
Page 2-11
Grady Chapter 3. Definitions Page 3-1
June 2006
3. Definitions
3.1. RMS
The squared rms value of a periodic current (or voltage) waveform is defined as
t o T
T
2 1 2
I rms
r s
T ³ i(t ) dt .
d (3.1)
to
It is clear in ((3.1)) that the squared rms value of a periodic waveform is the average value of the
squared waveform.
If the current is sinusoidal, the rms value is simply the peak value divided by
y 2 . However, if
the waveform has Fourier series
f
i (t ) ¦ I k sin(kZ1t T k ) ,
k 1
then substituting into (3.1) yields
to T 2
2 1 § f ·
I rms
T ³ ¨¨ ¦ I k sin(kZ1t T k ) ¸¸ dt
to © k 1 ¹
t T
1 o §¨ f 2 f f ·
2
I rms ³ ¦ I k sin 2
( kZ1t T k ) 2 ¦ ¦ I m I n sin( mZ1t T m ) sin(nZ1t T n ) ¸dt
T ¨ ¸
to © k 1 m 1, n 1,m z n ¹
t T
2 1 o §¨ f 2 § 1 cos 2( kZ1t T k ) ·
I rms ¦I ¨
T ³ ¨© k 1 k © 2
¸
¹
to
f f
§ cos((m n)Z1t T m T n ) cos((m n)Z1t T m T n ) · ·¸ (3.2)
¦ ¦ ImIn¨
© 2

2
¸ dt
¹ ¸¹
m 1, n 1,m z n
Equation (3.2) is complicated, but most of its terms contribute nothing to the rms value if one
thinks of (3.2) as being the average value for one fundamental period. The average value of each
cos 2(kZ1t T k ) term is zero because the average value of a cosine is zero for one or more
integer periods. Likewise, the average value of each cos((m r n)Z1t T m r T n ) term is also zero
because m and n are both harmonics of the fundamental. Thus, (3.2) reduces to
t T 2
2 1 o §¨ f 2 1 ·¸ 1 f 2 1 f 2 f
§I ·
I rm
rrms
ms ¦ I x dt
T ³ ¨© k 1 k 2 ¸¹
¦
2T k 1
I k x t o T t o ¦I
2k 1 k
¦ ¨¨ k2 ¸¸ , (3.3)
to k 1© ¹
Page 3-1
June 2006
where I k are peak values of the harmonic components. Factoring out the 2 yields
2
I rms I12, rms I 22,r 2
, rms I 3,r
, rms . (3.4)
Equations (3.3) and (3.4) ignore any DC that may be present. The effect of DC is to add the term
2
I DC to (3.3) and (3.4).
The cross products of unlike frequencies contribute nothing to the rms value of the total
waveform. The same statement can be made for average power, as will be shown later.
Furthermore, since the contribution of harmonics to rms add in squares, and their magnitudes are
often much smaller than the fundamental, the impact of harmonics on rms is usually not great.
3.2. THD
The most commonly-used measure for harmonics is total harmonic distortion (THD),
( ), also known
pp
as distortion factor. It is applied g and current. THD is defined as the rms value of
to both voltage
the harmonics above fundamental, divided by the rms values of the fundamental. DC is ignored.
Thus, for current,
f 2
§I · 1 f 2
¦ ¨¨ k2 ¸¸ ¦I
2k 2 k
k 2© ¹
THD
DI . (3.5)
I1 I1
2 2
The same equation form applies to voltage THDV .
THD and rms are directly linked. Note that since
2 1 f 2
I rms ¦I
2k 1 k
and since
1 f 2 § f 2· 2
¨ ¦ I ¸ I1
¦I
2k 2 k ¨ k ¸
THD I2 ©k 1 ¹ ,
I12 I12
2
then rewriting yields
Page 3-2
June 2006
f
¦ I k2
I12 1 THD I2 ,
k 1
so that
1 f 2 I12
¦I
2k 1 k 2

1 THD I2 .
Comparing to (3.3) yields
1 f 2 I12
2
I rms ¦I
2k 1 k 2

1 THD I2
I12, rms 1 THD I2 .
Thus the equation linking THD and rms is
2
I rms , rms 1 THD I .
I1,rms (3.6)
Because 1. line losses are proportional to the square of rms current ((and sometimes increase
more rapidly
p y due to the resistive skin effect),
), and 2. rms increases with harmonics, then line
losses always increase when harmonics are present. For example, many PCs have a current
distortion near 1.0 (i.e., 100%). Thus, the wiring losses incurred while supplying a PC are twice
what they would be in the sinusoidal case.
p
Current distortion in loads varies from a few percent to more than 100%,, but voltage
g distortion is
g
generallyy less than 5%. Voltage
g THDs below 0.05,, i.e. 5%,, are considered acceptable, and those
greater than 10% are definitely unacceptable and will cause problems for sensitive equipment
and loads.
3.3. Average Power
Harmonic powers (including the fundamental) add and subtract independently to produce total
average power. Average power is defined as
t o T to T
1 1
Pavg
a
T ³ pp(t)
(t )dt
T ³ v(t )i(t )dt . (3.7)
to to
Substituting in the Fourier series of voltage and current yields
to T
1 § f f ·
¨ ¦ Vk sin( kZ1t G k ) x ¦ I k sin( kZ1t T k ) ¸dt ,
Pavg
T ³ ¨ ¸
to © k 1 k 1 ¹
Page 3-3
June 2006
and expanding yields
t T
1 o §¨ f
Pavg ¦ Vk I k sin(kZ1t I k ) x sin(kZ1t T k )
T ³ ¨© k 1
to
f f ·
¸ ,
¦ ¦ mn V I sin( mZ1t I m ) sin( nZ1t T n ¸dt
)
m 1, n 1,m z n ¹
t T
1 o §¨ f § cos(I k T k ) cos(2kZ1t I k T k ) ·
Pavg
T ³ ¨ ¦ Vk I k ¨
© 2
¸
¹
t ©k 1
o
f f
§ cos((m n)Z1t T m T n ) cos((m n)Z1t T m T n ) · ·¸
¦ ¦ ImIn¨
© 2

2
¸ dt
¹ ¸¹
m 1, n 1,m z n
As observed in the rms case, the average value off all the sinusoidal term
termss is zero, leaving only
the time invariant terms in the summation, or
f Vk I k f
Paavg
vg ¦ cos(I k T k ) ¦ Vk ,rms x I k ,rms x dpf k a P22,,avg P33,,avg
P1,avg , (3.8)
k 1 2 k 1
where dpf k is the displacement power factor for harmonic k .
h
The harmonic power terms P2,,avg , avg , are mostly losses and are usually small in relation
avg , P3,avg
to total power. However, harmonic losses may be a substantial part of total losses.
Equation (3.8) is important in explaining who is responsible for harmonic power. Electric utility
generating
g g plants
p produce
p sinusoidal terminal voltages.
g Accordingg to (3.8),
( ), if there is no
g
harmonic voltage at the terminals of a generator,, then the generator
g pproduces no harmonic
power.
p However,, due to nonlinear loads, harmonic power does indeed exist in power systems
and causes additional losses. Thus, it is accurate to say that
x Harmonic power is parasitic and is due to nonlinear equipment and loads.
x The source of most harmonic powerr is power electronic loads.
x Byy chopping
pp g the 60 Hz current waveform and pproducing g harmonic voltages
g and currents,
p
power electronic loads convertt some of the “60 Hz” power
p into harmonic power, which
in turn propagates back into the power system, increasing system losses and impacting
sensitive loads.
For a thought provoking question related to harmonic power, consider the case shown in Figure
3.1 where a perfect 120Vac(rms) power system with 1ȍ internal resistance supplies a triac-
Page 3-4
June 2006
controlled 1000W incandescent lamp. Let the firing angle is 90°, so the lamp is operating at
half-power.
Current i
1ȍ
i
+ +
vs vm Customer 14.4ȍ
Wattmeter
– here –
Figure 3.1. Single-Phase Circuit with Triac and Lamp
120 2
Assuming that the resistance of the lamp is = 14.4ȍ, and that the voltage source is
1000
v s (t ) 120 2 sin(Z1t ) , then the Fourier series of current in the circuit, truncated at the 5th
harmonic, is
i (t ) 6.99 sin(Z1t 32.5 o ) 3.75 sin(3Z1t 90.0 o ) 1.25 sin(5Z1t 90.0 o ) .
If a wattmeter is placed immediately to the left of the triac, the metered voltage is
v m (t ) v s (t ) iR 120 2 sin(Z1t )

1 x 6.99 sin(Z1t 32.5 o ) 3.75 sin(3Z1t 90.0 o ) 1.25 sin(5Z1t 90.0 o )
163.8 sin(Z1t 1.3o ) 3.75 sin(3Z1t 90.0 o ) 1.25 sin(5Z1t 90.0 o )
and the average power flowing into the triac-lamp customer is
Pavg
163.8 x 6.99
2

cos 1.3o (32.5 o )
3.75 x 3.75
2

cos 90.0 o (90.0 o )

1.25 x 1.25
2

cos 90.0 o (90.0 o )
Page 3-5
June 2006
= 475.7 – 7.03 – 0.78 = 467.9W
The first term,, 475.7W,, is due to the fundamental component of voltage and current. The 7.03W
and 0.78W terms are due to the 3rdd and 5th harmonics, respectively, and flow back into the power
system.
The question now is: should the wattmeter register only the fundamental power, i.e., 475.7W, or
should the wattmeter credit the harmonic power flowing back into the power system and register
only 475.7 – 7.81 = 467.9W? Remember that the harmonic power produced by the load is
consumed by the power system resistance.
3.4. True Power Factor
To examine the impact of harmonics on power factor, it is important to consider the true power
factor, which is defined as
Pavg
a
pff true
t . (3.9)
Vrms
r s I rrms
In sinusoidal situations, (3.9) reduces to the familiar displacement power factor

V1 I1
P1, avg cos(G 1 T1 )
dpff1 2 cos((G 1 T1 ) .
V1, rms I1, rms V1 I1
2
When harmonics are present, (3.9) can be expanded as
P1, avg P2, avg P3, avg

pf true .
V1, rms 1 THDV2 x I1, rms 1 THD I2
In most instances, the harmonic powers are small compared to the fundamental power,
p , and the
voltage distortion is less than 10%. Thus, the following important simplification is usually valid:
P1, avg dpf1

pff ttrue | . (3.10)
V1, rms I1, rms 1 THD I2 D I2
1 THD
It is obvious in (3.10) that the true power factor of a nonlinear load is limited by its THD I . For
example, the true power factor of a PC with THD I = 100% can never exceed 0.707, no matter
how good its displacement power is. Some other examples of “maximum” true power factor
(i.e., maximum implies that the displacement power factor is unity) are given below in Table 3.1.
Page 3-6
June 2006
Table 3.1. Maximum True Power Factor of a Nonlinear Load.
Current Maximum
THD pf true
20% 0.98
50% 0.89
100% 0.71
3.5. K Factor
Losses in transformers increase when harmonics are present because
1. harmonic currents increase the rms current beyond what is needed to provide load power,
2. harmonic currents do not flow uniformly throughout the cross sectional area of a
conductor and thereby increase its equivalent resistance.
Dry-type transformers are especially sensitive to harmonics. The K factor was developed to
provide a convenient measure for rating the capability of transformers, especially dry types, to
serve distorting loads without overheating. The K factor formula is
f
¦ k 2 I k2
k 1 . (3.11)
K
f
¦ I k2
k 1
In most situations, K d 10 .
3.6. Phase Shift
There are two types of phase shifts pertinent to harmonics. The first is a shift in time, e.g. the
2T
r among the phases of balanced a-b-c currents. To examine time shift, consider Figure 3.2.
3
If the PC waveform is delayed by 'T seconds, the modified current is
f f
i (t 'T ) ¦ I k sin(kZ1 t 'T T k ) = ¦ I k sin(kZ1t kZ1'T T k )
k 1 k 1
f f
¦ I k sin kZ1t T k kZ1'T = ¦ I k sinkZ1t T k kT1 , (3.12)
k 1 k 1
where T1 is the phase lag of the fundamental current corresponding to 'T . The last term in
(3.12) shows that individual harmonics are delayed by kT1 .
Page 3-7
June 2006
5
delayed
Amperes
0
-5
Figure 3.2. PC Current Waveform Delayed in Time
The second type of phase shift is in harmonic angle, which occurs in wye-delta transformers.
Wye-delta transformers shift voltages and currents by r 30 o . ANSI standards require that,
regardless of which side is delta or wye, the a-b-c phases must be marked so that the high-
voltage side voltages and currents lead those on the low-voltage side by 30 o for positive-
sequence, andd lag by 30 o for negative sequence. Zero sequences are blocked by the three-wire
connection so that their phase shift is not meaningful.
3.7. Voltage and Current Phasors in a Three-Phase System
Phasor diagrams for line-to-neutral and line-to-line voltages are shown in Figure 3.3. Phasor
currents for a delta-connected load, and their relationship to line currents, are shown in Figure
3.4.
Page 3-8
June 2006
Imaginary
Vca = Vcn – Van Vcn Vab = Van – Vbn
30°
Real
Van
120°
Vbn
Vbc =
Vbn – Vcn
Figure 3.3. Voltage Phasors in a Balanced Three-Phase System

(The phasors are rotating counter-clockwise. The magnitude of line-to-line voltage phasors is 3 times the magnitude
of line-to-neutral voltage phasors.)
Page 3-9
June 2006
Imaginary
Vca = Vcn – Van Vcn Vab = Van – Vbn
Ic
Ica
30°
Iab
Real
Ia Van
Line currents Ia, Ib, and Ic
Ib Ibc Delta currents Iab, Ibc, and Ica
Ic c
Vbn
Ica
Balanced Sets Add to Zero in Both Ibc

Time and Phasor Domains
I a + Ib + I c = 0
Iab
Ib b
Van + Vbn + Vcn = 0 a
Vbc =
Vbn – Vcn – Vab +
Vab + Vbc + Vca = 0
Ia
Figure 3.4. Currents in a Delta-Connected Load
1
(Conservation of power requires that the magnitudes of delta currents Iab, Ica, and Ibc are times the magnitude of line currents Ia, Ib, Ic.)
3
Page 3-10
June 2006
3.8. Phase Sequence
In a balanced three-phase power system, the currents in phases a-b-c are shifted in time by
r 120 o of fundamental. Therefore, since
f
ia (t ) ¦ I k sin(kZ1t T k ) ,
k 1
2S
then the currents in phases b and c lag and lead by radians, respectively. Thus
3
f
2S
ib (t ) ¦ I k sin(kZ1t T k k 3
),
k 1
f
2S
ic (t ) ¦ I k sin(kZ1t T k k 3
).
k 1
Picking out the first three harmonics shows an important pattern. Expanding the above series,
ia (t ) I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 ) ,
2S 4S 6S
ib (t ) I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 ) , or
3 3 3
2S 2S
I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 0) .
3 3
2S 4S 6S
ic (t ) I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 ) , or
3 3 3
2S 2S
I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 0) .
3 3
By examining the current equations, it can be seen that
x the first harmonic (i.e., the fundamental) is positive sequence (a-b-c) because phase b lags
phase a by 120º, and phase c leads phase a by 120º,
x the second harmonic is negative sequence (a-c-b) because phase b leads phase a by 120º,
and phase c lags phase a by 120º,
x the third harmonic is zero sequence because all three phases have the same phase angle.
Page 3-11
June 2006
The pattern for a balanced system repeats and is shown in Table 2. All harmonic multiples of
three (i.e., the “triplens”) are zero sequence. The next harmonic above a triplen is positive
sequence, the next harmonic below a triplen is negative sequence.
Table 3.2. Phase Sequence of Harmonics in a Balanced Three-Phase System
Phase
Harmonic Sequence
1 +
2 –
3 0
4 +
5 –
6 0
… …
If a system
y is not balanced,, then each harmonic can have ppositive,, negative,
g , and zero sequence
q
components. However, in most cases, the pattern in Table 3.2 can be assumed to be valid.
Because of Kirchhoff’s current law, zero sequence currents cannot flow into a three-wire
connection such as a delta transformer winding or a delta connected load. In most cases, systems
are fairly well balanced,, so that it is common to make the same assumption
p for third harmonics
p
and other triplens. Thus,, a delta-grounded
g wye
y transformer at the entrance of an industrial
customer usually y blocks the flow of triplen
p harmonic load currents into the ppower system.
y
Unfortunately, the transformer does nothing to block the flow of any other harmonics, such as 5th
and 7th.
q
Zero sequence g neutral or grounding
currents flow through g paths. Positive and negative sequence
currents sum to zero at neutral and grounding points.
Another interesting observation can be made about zero sequence harmonics. Line-to-line
voltages never have zero sequence components because, according to Kirchhoff’s voltage law,
they always sum to zero. For that reason, line-to-line voltages in commercial buildings are
missing the 3rd harmonic that dominates line-to-neutral voltage waveforms. Thus, the THDV of
line-to-line voltages is often considerably less than for line-to-neutral voltages.
3.9. Transformers
Consider the example shown in Figure 3.5 where twin, idealized six-pulse current source ASDs
are served by parallel transformers. Line-to-line transformer voltage ratios are identical. The top
transformer is wye-wye or delta-delta, thus having no phase shift. The bottom transformer is
wye-delta or delta-wye, thus having 30 o phase shift.
Page 3-12
June 2006
Top,
loadside
THD I = 30.0%
ǻǻ
Six-Pulse
or ASD
YY
THD I = 14.2%
Net Twelve-Pulse
Ȉ line-side load-side
Operation ǻY
Six-Pulse
or ASD
Yǻ
THD I = 30.0% THD I = 30.0%
Figure 3.5. Current Waveforms of Identical Parallel Six-Pulse Converters

Yield a Net Twelve-Pulse Converter
g the analysis,
To begin y , assume that the per-unit load-side current of the top transformer is the
standard six-pulse wave given by
itopp, loadside (t ) I1 sin(1Z1t )

I I
sin(5Z o t 180 o ) 1 sin(
1 sin(5 sin(77Z1t 180 o )
5 7
I I
1 sin(111Z1t ) 1 sin(133Z1t )
11 13
I1 I
sin(177Z1t 180 o ) 1 sin(19 9Z1t 180 o )
17 19
Note that the even-ordered harmonics are missingg because of half-wave symmetry,
y y, and that the
triple harmonics are missing because a six-pulse ASD is a three-wire balanced load, having
Page 3-13
June 2006
characteristic harmonics k 6n 6n r 1, n 1,2,3,... . Because the transformer has no phase shift,
then the line-side current waveform (in per-unit) is the same as the load-side current, or
itop , lineside (t ) itop , loadside (t ) .
Now, because the fundamental voltage on the load-side of the bottom transformer is delayed in
time by 30 o , then each harmonic of the load-side curr
current
ent of the bottom transformer is delayed by
30 o , so that
k x3
ibottom, loadside (t ) I1 sin(1Z1t 30 o )

I I
1 sin(5Z1t 180 o 150 o ) 1 sin(7Z1t 180 o 210 o )
5 7
I I
1 sin(11Z1t 330 o ) 1 sin(13Z1t 390 o )
11 13
I I
1 sin(17Z1t 180 o 510 o ) 1 sin(19Z1t 180 o 570 o )
17 19
The current waveform through the top transformer is not shifted when going from load-side to
line-side, except for its magnitude. However, the various phase sequence components of the
current through the bottom transformer are shifted when going to the line-side, so that
ibottom, lineside (t ) I1 sin(1Z1t 30 o 30 o )

I I
1 sin(5Z1t 180 o 150 o 30 o ) 1 sin(7Z1t 180 o 210 o 30 o )
5 7
I I
1 sin(11Z1t 330 o 30 o ) 1 sin(13Z1t 390 o 30 o )
11 13
I1 I
sin(17Z1t 180 o 510 o 30 o ) 1 sin(19Z1t 180 o 570 o 30 o ) .
17 19
Combining angles yields
ibottom, lineside (t ) I1 sin(1Z1t )

I I
1 sin(5Z1t ) 1 sin(7Z1t )
5 7
I1 I
sin(11Z1t ) 1 sin(13Z1t )
11 13
I I
1 sin(17Z1t 180 o ) 1 sin(19Z1t ) .
17 19
Adding the top and bottom line-side currents yields
Page 3-14
June 2006
2 I1 2I
inett (t
(t ) (t ) ibottom, linesidee (t
itop , linesidee (t (t ) 2 I1 ssin(1Z1t ) sin(111Z1t ) 1 sin(133Z1t ) .
11 13
p
The important observation here is that harmonics 5,7,17,19 combine to zero at the summing
point on the line-side. Recognizing the pattern shows that the remaining harmonics are
k 12n r 1, n 1,2,3,... ,
which leads to the classification “twelve-pulse converter.”
In an actual twelve-pulse ASD, a three winding transformer is used, having one winding on the
line-side, and two parallel wye-delta and delta-delta windings on the load-side. The power
electronics are in effect divided into two halves so that each half carries one-half of the load
power.
Summarizing, since harmonics in a balanced system fall into the predictable phase sequences
shown in Table 3.2, it is clear that a wye-delta transformer will advance some harmonics by 30 o
and delayy other harmonics byy 30 o . This pproperty
p y makes it ppossi
possible
ble to cancel half of the
th th
harmonics produced
p by ASDs (most importantly the 5 and 7 ) through a principle known as
phase cancellation. The result is illustrated in Figure 3.3, where two parallel six-pulse converters
combine to yield a net twelve-pulse converter with much less current distortion. Corresponding
spectra are given in the Appendix.
Transformer phase shifting may be used to create net 18-pulse, 24-pulse, and higher-pulse
converters.
Page 3-15
Grady Chapter 4. Sources Page 4-1
June 2006
4. Sources
Harmonics are produced by nonlinear loads or devices that draw non no sinusoidal currents. An
example of a nonlinear load is a diode, which permits only one-half of the otherwise sinusoidal
current to flow. Another example is a saturated transformer, whose magnetizing current is no
sinusoidal. But, by far the most common problem-causing nonlinear loads are large rectifiers
and ASDs.
Nonlinear load current waveshapes always vary somewhat with the applied voltage waveshape.
Typically, the current distortion of a nonlinear load decreases as the applied voltage distortion
increases – thus somewhat of a compensating effect. As a result, most nonlinear loads have the
highest current distortion when the voltage is nearly sinusoidal and the connected power system
is “stiff” (i.e., low impedance).
In most harmonics simulation cases, these waveshape variations are ignored and nonlinear loads
are treated as fixed harmonic current injectors whose harmonic current magnitudes and phase
angles are fixed relative to their fundamental current magnitude and angle. In other words, the
harmonic current spectrum of a nonlinear load is usually assumed to be fixed in system
simulation studies. The fundamental current angle, which is almost always lagging, is adjusted
to yield the desired displacement power factor. Harmonics phase angles are adjusted according
to the time shift principle to preserve waveshape appearance.
4.1 Classical Nonlinear Loads
Some harmonic sources are not related to power electronics and have been in existence for many
years. Examples are
x Transformers. For economic reasons, power transformers are designed g p

to operate on or
g y past
slightly p the knee of the core material saturation curve. The resulting magnetizing
current is slightly peaked and rich in harmonics. cs The third harmonic component
dominates. Fortunately, magnetizing current is only a few percent of full-load current.
The magnetizing current for a 25 kVA, 12.5kV/240V transformer is shown in Figure 4.1
(see spectrum in the Appendix). The fundamental current component lags the
g component
fundamental voltage p y 66q. Even though
by g the 1.54Arms magnetizing
g g current
g y distorted (76.1%), it is relatively small compared to the rated full-load current of
is highly
140Arms.
Page 4-1
June 2006
Amperes
0
-2
-4
-6
Figure 4.1. Magnetizing Current for Single-Phase

g 25 kVA. 12.5kV/240V Transformer.
THD I = 76.1%.
x Machines. As with transformers,, machines operate

p with ppeak flux densities beyond
y the
saturation knee. Unless blocked byy a delta transformation, a three-phase synchronous
generator will produce a 30% third harmonic current.
There is considerable variation among single-phase motors in the amount of current

harmonics they inject. Most of them have THD I in the 10% range, dominated by the 3rd
harmonic. The current waveforms for a refrigerator and residential air conditioner are
shown in Figures 4.2 and 4.3, respectively. The corresponding spectra are given in the
Appendix. The current waveform for a 2HP single-phase motor is shown in Figure 5.5 in
Section 5.
2
Amperes
-2
-4
Figure 4.2. 120V Refrigerator Current.

THD I = 6.3%.
Page 4-2
June 2006
40
30
20
10
Amperes
0
-10
-20
-30
-40
Figure 4.3. 240V Residential Air Conditioner Current.

THD I = 10.5%.
x Fluorescent Lamps (with Magnetic Ballasts). Fluorescent lamps extinguish and ignite
each half-cycle, but the flicker is hardly perceptable at 50 or 60Hz. Ignition occurs
sometime after the zero crossing of voltage. Once ignited, fluorescent lamps exhibit
negative resistive characteristics. Their current waveforms are slightly skewed, peaked,
and have a characteristic second peak. The dominant harmonics is the 3rd, on the order of
15% - 20% of fundamental. A typical waveform is shown in Figure 4.4, and the
spectrum is given in the Appendix.
0.3
0.2
0.1
Amperes
0
-0.1
-0.2
-0.3
Figure 4.4. 277V Fluorescent Lamp

p Current (with Magnetic Ballast).
THD I = 18.5%.
x Arc Furnaces. These are not strictly periodic and, therefore, cannot be analyzed
accurately by using Fourier series and harmonics. Actually, these are transient loads for
which flicker is a greater problem than harmonics. Some attempts have been made to
model arc furnaces as harmonic sources using predominant harmonics 3rd and 5th.
4.2. Power Electronic Loads
Examples of power electronic loads are
Page 4-3
June 2006
x Line Commutated Converters. These are the workhorse circuits of AC/DC converters
above 500HP. The circuit is shown in Figure 4.5. These are sometimes described as six-
pulse converters because they produce six ripple peaks on Vdc per AC cycle. In most
applications, power flows to the DC load. However, if the DC circuit has a source of
power, such as a battery or photovoltaic array, power can flow from DC to AC in the
inverter mode.
The DC choke smooths Idc, and since Idc has low ripple, the converter is often described
as a “current source.”
In order to control power flow, each SCR is fired after its natural forward-bias turn-on
point. This principle is known as phase control, and because of it, the displacement
power factor is poor at medium and low power levels.
The firing order is identified by SCRs 1 through 6 in Figure 4.5. Once fired, each SCR
conducts until it is naturally reverse biased by the circuit. The term “line commutated
converter” refers to the fact that the load actually turns the SCRs off, rather than having
forced-commutated circuits turn them off. Line commutation has the advantage of
simplicity.
The idealized AC current ia (t ) waveform for a six-pulse converter equals Idc for 120º,
zero for 60º, and then –Idc for 120º, and zero for another 60º (see Figure 4.5 and the field
measurement shown in Figure 5.1). The Fourier series is approximately
1 1
i (t ) I1 > sin(1Z1t 1T1 ) sin(5Z1t 5T1 ) sin(7Z1t 7T1 )
5 7
1 1
sin(11Z1t 11T1 ) sin(13Z1t 13T1 )
11 13
1 1
sin(17Z1t 17T1 ) sin(19Z1t 19T1 ) @ ,
17 19
where I1 is the peak fundamental current, and T1 is the lagging displacement power
factor angle.
g The magnitudes
g of the AC current harmonics decrease by y the 1/k rule,, i.e.
the fifth harmonic is 1/5 of ffundamental,, the seventh harmonic is 1/7 of fundamental,, etc.
The even-ordered harmonics are missingg due to half-wave symmetry,y y, and the triple
p
harmonics are missingg because the converter is a three-wire load served by a transformer
with a delta or ungrounded-wye winding.
Page 4-4
June 2006
ia’ waveform. THD I = 30.0%.
i1 Idc
+ Choke
1 3 5
ia ia’
a DC Load
b Vdc or
c Inverter
Transformer
4 6 2
–
Idc
ia with delta-wye or wye-delta transformer. ia with delta-delta or wye-wye transformer.

THD I = 30.0% in both cases.
Figure 4.5. Three-Phase, Six-Pulse Line Commutated Converter
Page 4-5
June 2006
If the converter transformer has no phase shift (i.e., either wye-wye or delta-delta), then
the current waveshape on the power system side, i.e., ia (t ) , is the same as current ia' (t )
on the converter side of the transformer. If the transformer is wye-delta or delta-wye,
then the sign of every other pair of harmonics in ia (t ) changes, yielding
1 1
ia (t ) >
I1 sin(1Z1t 1T1 ) sin(5Z1t 5T1 ) sin(7Z1t 7T1 )
5 7
1 1
sin(11Z1t 11T1 ) sin(13Z1t 13T1 )
11 13
1 1
sin(17Z1t 17T1 ) sin(19Z1t 19T1 ) @ .
17 19
Two or more six-pulse converters can be operated in parallel through phase-shifting

transformers to reduce the harmonic content of the net supply-side current. This principle
is known as phase cancellation. A twelve-pulse converter has two six-pulse converters
connected in parallel on the AC side and in series on the DC side. One load-side
transformer winding is delta and the other is wye. As a result, half of the harmonic
currents cancel (notably, the 5th and 7th), producing an AC current waveform that is
much more sinusoidal than that of each individual converter alone. Higher pulse orders
(i.e., eighteen pulse, twenty-four pulse, etc.) can also be achieved. The AC current
harmonic multiples produced by a P-pulse converter are
h = PN ± 1, N = 1,2,3, ... ,
P = an integer multiple of 6.
x Voltage-Source Converters. For applications below 500HP, voltage source converters

employing pulse-width modulaters with turn-on/turn-off switches on the motor side are
often the choice for ASDs. Since both power and voltage control is accomplished on the
load side, the SCRs in Figure 4.5 can be replaced with simple diodes. The circuit is
shown in Figure 4.6, and the spectra are given in the Appendix.
The diode bridge and capacitor provide a relatively stiff Vdc source for the PWM drive,
hence the term “voltage source.” Since voltage-source converters do not employ phase
control, their displacement power factors are approximately 1.0.
Page 4-6
June 2006
ia with high power. THD I = 32.6%. ia with low power. THD I = 67.4%.
(delta-delta or wye-wye) (delta-delta or wye-wye)
i1
+
1 3 5
ia
a DC Load
b or
Vdc
c PWM Inverter
Transformer
Capacitor
4 6 2
–
ia with high power. THD

D I = 32.6%. ia with low power. THD
D I = 67.4%.
(delta-wye or wye-delta) (delta-wye or wye-delta)
Figure 4.6. Three-Phase, Six-Pulse Voltage-Source Converter
Page 4-7
June 2006
Unfortunately, current distortion on the power system side is higher for voltage-source
converters than for line commutated converters, and the current waveshape varies
considerably with load level. Typical waveforms are shown in Figure 4.6. Even though
lower load levels have higher THD I , the harmonic amperes do not vary greatly with load
level because fundamental current is proportional to load level.
The higher
g current distortion created byy these drives is one of the main reasons that
voltage-source inverters are generally not used above 500HP.
x Switched-Mode Power Supplies. These power supplies are the "front-end" of single-
phase 120V loads such as PCs and home entertainment equipment. Typically, they have
a full-wave diode rectifier connected between the AC supply system and a capacitor, and
the capacitor serves as a low-ripple “battery” for the DC load. Unfortunately, low ripple
means that the AC system charges the capacitor for only a fraction of each half-cycle,
yielding an AC waveform that is highly peaked, as shown in Figure 4.7.
Page 4-8
June 2006
i1
+
1 3
i
DC Load
Or DC/DC
Vdc
Converter
Capacitor
4 2
–
Waveform for one PC

Amperes
-5
AC Current for Above Circuit. THD

D I = 134%.
With one PC on each
phase, the 3rd harmonic
5
current is missing on the
delta side of a delta-
grounded wye
Amperes
transformer, and the

positive and negative
0 harmonics are phase
shifted. The time domain
waveform changes
dramatically.
-5
AC Current on Delta Side of Delta-Grounded

d Wye Transformer that Serves Three PCs.
THD I = 94.0%.
Figure 4.7. Single-Phase Switched-Mode Power Supply and Current Waveforms
Page 4-9
June 2006
4.3. Other Nonlinear Loads
There are many other harmonic sources. Among these are cycloconverters, which directly
convert 60 Hz AC to another frequency, static VAr compensators, which provide a variable
supply of reactive power, and almost any type of "energy saving" or wave-shaping device, such
as motor power factor controllers. Waveforms for three common loads are shown below in
Figures 4.8, 4.9, and 4.10, and the corresponding spectra are given in the Appendix.
25
20
15
10
Amperes
5
0
-5
-10
-15
-20
-25
Figure 4.8. 120V Microwave Oven Current.

THD
D I = 31.9%.
12
8
4
Amperes
0
-4
-8
-12
Figure 4.9. 120V Vacuum Cleaner Current.

THDD I = 25.9%.
Page 4-10
June 2006
0.2
0.15
0.1
0.05
Amperes
0
-0.05
-0.1
-0.15
-0.2
Figure 4.10. 277V Fluorescent Lamp Current (with Electronic Ballast).

THDD I = 11.6%.
4.4. Cumulative Harmonics
Voltage
g distortion and load level affect the current waveshapes p of nonlinear loads. Harmonic
magnitudes
g p
and phase g , especially
angles, p y the phase
p angles of higher-frequency harmonics, are a
function of waveshape and displacement power factor. Thus, the net harmonic currents produced
by ten or more nearby harmonic loads are not strictly additive because there is some naturally-
occuring phase cancellation. If this phase angle diversity is ignored, then system simulations will
predict exaggerated voltage distortion levels.
This net addition, or diversity factor, is unity for the 3rd harmonic, but decreases for higher
harmonics. Research and field measurement verifications have shown that the diversity factors
in Table 4.1 are appropriate in both three-phase and single-phase studies. Even-ordered
harmonics are ignored.
Table 4.1. Current Diversity Factor Multipliers for Large Numbers of Nonlinear Loads
Current Diversity
Harmonic Factor
3 1.0
5 0.9
7 0.9
9 0.6
11 0.6
13 0.6
15 0.5
Higher
g Odds 0.2
All Evens 0.0
A typical application of Table 4.1 is when ten 100HP voltage-source ASDs are located within a
single facility, and the facility is to be modeled as a single load point on a distribution feeder.
The net ASD is 1000HP, and the net spectrum is the high-power spectrum of Figure 4.6 but with
Page 4-11
June 2006
magnitudes multiplied by the diversity factors of Table 4.1. The phase angles are unchanged.
The composite waveshape is shown in Figure 4.11.
Current distortion for one

drive is 32.6%, but for
combined drives is
27.6%, with high-order
harmonics significantly
reduced.
Figure 4.11. Expected Composite Current Waveshape for Large Numbers of High-Power
Voltage-Source ASDs. THD I = 27.6%.
Similiarly, the composite waveshape for one thousand 100W PCs with the waveform shown in
Figure 4.7 would be a single 100kW load with the waveshape shown in Figure 4.12.
High-order harmonics
significantly reduced, but
distortion still high
because the 3rd harmonic
was not affected.
Figure 4.12. Expected Composite Current Waveshape for Large Numbers of PCs.
THD I = 124%.
Page 4-12
June 2006
4.5. Detailed Analysis
y of Steady-State
y Operation
p of Three-Phase, Six-Pulse, Line
Commutated, Current-Source Converters
p
Ripple-free
Idc
+
–
V1 1 3 5 The converter
L + i1 i3 i5 is connected to
Van ia a DC circuit,
L
Vbn ib which consists
L Vdc
ic of a Thevenin
Vcn
equivalent R
and V in series
with a large
Net transformer and
4 6 2 inductor.
system inductance
i4 i6 i2
Idc
Figure 4.13. Three-Phase, Six-Pulse, Line Commutated, Current-Source Converter
Introduction
Line-commutated converters are most-often used in high-power
g p applications
pp such as motor
drives (larger than a few hundred kW) and HVDC (hundreds of MW). These applications
require the high voltage and current ratings
g that are ggenerallyy available only in thyristors (i.e.,
silicon-controlled rectifiers, or SCRs). A large series inductor is pplaced in the DC circuit to
lower the ripple content of Idc,, which in turn helps to limit the harmonic distortion in the AC
currents to approximately 25%.
Byy adjusting
j ngle Į, the converter can send ppower from the AC side to the DC side (i.e.,
g firingg angle
an
rectifier operation), or from the DC side to the AC side (i.e., inverter operation). DC voltage
Vdc is positive for rectifier operation, and negative for inverter operation. Because thyristors are
unidirectional, DC current always flows in the direction shown.
To understand the operating principles, the following assumptions are commonly made
x Continuous and ripple free Idc
x Balanced AC voltages and currents
x Inductive AC system impedance
x Balanced, steady-state operation with
x firing angle Į, 0˚ Į 180˚,
x commutation angle ȝ, 0˚ ȝ 60˚, and
x 0˚ Į + ȝ 180˚.
Page 4-13
June 2006
As a first approximation,
pp when Į < 90˚,, then the circuit is a rectifier. When Į > 90˚,, then the
circuit is an inverter. The zero reference for Į is the point at which turn-on would naturally
occur if a thyristor was replaced by a diode.
The waveform ggraphs

p shown in this document are pproduced by Excel spreadsheets
6P_Waveforms_Rectifier.XLS and 6PLCC_Waveforms.XLS.
Simple Uncontrolled Rectifier with Resistive Load

A good starting point for understanding the operation of the converter is to consider the circuit
shown in Figure 4.14, where the thyristors have been replaced with diodes, the DC circuit is
simply a load resistor, and the AC impedance is negligible. Without a large inductor in the DC
circuit, the DC current is not ripple-free.
Idc
Using KCL here,
ia = i1 – i 4 – 1 3 5 +
V1
+ i1 i3 i5
Van ia
Vbn ib
ic Vdc Rload
Vcn
4 6 2
i4 i6 i2 –
Idc
Figure 4.14. Three-Phase Uncontrolled Rectifier
The switching rules for Figure 4.14 are described below.

x Diode #1 is on when Van is the most positive, i.e., Van > Vbn and Van > Vcn.
Simultaneously, diodes #3 and #5 are reverse biased and thus off.
x Likewise, diode #4 is on when Van is the most negative. Simultaneously, diodes #2 and
#6 are reverse biased and thus off.
x The other diodes work in the same way that #1 and #4 do.
x At any time, one (and only one) of top diodes #1 or #3 or #5 is on, and one (and only
one) of bottom diodes #2 or #4 or #6 is on.
x The pair of top and bottom diodes that is on determines which line-to-line voltage
appears at Vdc.
Using the above rules, waveforms for i1, … , i6, ia, V1, and Vdc can be determined and are
shown in Figure 4.15. The figure confirms the natural turn-on sequence for diodes #1, #2, … #6.
Page 4-14
June 2006
Figure 4.15. Waveforms for the Three-Phase Uncontrolled Rectifier with Resistive Load
(note – the graph contains the phrase “uncontrolled rectifier,” but when Į = 0°, controlled and
uncontrolled rectifiers are essentially the same)
Page 4-15
June 2006
Simple Controlled Rectifier with Resistive Load
Now, replace the diodes in Figure 4.14 with SCRs so that power can be controlled. When fired,
SCRs will turn on if they are forward biased. The point at which they first become forward
biased corresponds to a firing angle Į of 0° – that is the same situation as the diode case of
Figure 4.14. If firing angle Į = 30°, then firing occurs 30° past the point at which the SCRs first
become forward biased.
Working with the switching rules given for Figure 4.14, and modifying them for Į > 0, the
waveforms can be determined and are shown in Figure 4.16.
At this point, it should be noted that if Į is greater than 60°, then the load current becomes
discontinuous.
Page 4-16
June 2006
30 degrees
Figure 4.16. Waveforms for the Three-Phase Controlled Rectifier with Resistive Load
Page 4-17
June 2006
Controlled Operation of Three-Phase, Six-Pulse, Line Commutated,
Current-Source Converter
We now return to the circuit shown in Figure
g 4.13. The DC circuit has a smoothingg inductor to
remove ripple,
pp , and the AC system
y g
has an inductance. The significance of the AC inductance
means that when SCR #1 turns on, SCR #5 does not immediately turn off. Gradually, Idc
transitions from SCR #5 to SCR #1. This transition is known as commutation. In industrial
converters, commutation angle μ may be only 2-3° of 60 Hz. In HVDC converters, commutation
may be intentionally increased to 10-15° to reduce AC harmonics.
To understand circuit operation, consider the 60° sequence where

Case 1. #5 and #6 are on (just prior to #1 being fired),
Case 2. Then #1 is fired, so that #1 and #5 commutate (while #6 stays on),
Case 3. Then #5 goes off, and #1 and #6 are on.
Once this 60° sequence is understood, then because of symmetry, the firing of the other SCRs
and their waveforms are also understood for the other 300° that completes one cycle of 60Hz.
Vcn
#5 on
–
Vac +
L +
Van ia
L
Vbn ib Idc
L Vcb
Vcn ic Idc
Zt 30 o D #6 on –
Vbn
Case 1. #5 and #6 On (just prior to firing #1)
Page 4-18
June 2006
Idc 1
(Van+Vcn)
#1 on #5 on 2
–
0 +
L +
Van ia i1
L 3
Vbn ib Idc Vbn
L 2
Vcn ic i5
–
o o #6 on
30 D d Zt d 30 D P
Vbn
Case 2. #1 and #5 Commutating (#6 stays on),
Van
#1 on
–
0 +
L +
Van ia Idc
L
Vbn ib Idc
L Vab
Vcn ic
–
30 o D P d Zt d 90 o D #6 on
Vbn
Case 3. #1 and #6 On
The analysis for commutation in Case 2 follows. When #1 comes on, KVL around the loop
created by #1, #5, and Vac yields
di di
Vac L 1 L 5 0.
dt dt
KCL at the top DC node yields
i1 i5 I dc 0 , so that i5 I dc i1 .
Page 4-19
June 2006
Substituting the KCL equation into the KVL equation yields
di d ( I dc i1 )
Vac L 1 L 0.
dt dt
Since I dc is constant, then
di d (i1 ) di d (i )
Vac L 1 L Vac L 1 L 1 0 , which becomes
dt dt dt dt
di1 Vac
.
dt 2L
Thus
Vac V LLP V LLP
i1 ³ 2 L dt ³ sin(Zt 30 o )dt cos(Zt 30 o ) const .
2L 2ZL
The boundary conditions are i1 (Zt 30 o D ) 0 . Therefore,
V LLP
const cos(D ) , so that
2ZL
i1
V LLP
2ZL

cos(D ) cos(Zt 30 o ) , 30 o D d Zt d 30 o D P . (4.1)
From the KCL equation,
i5 I dc i1
V
2ZL

I dc LLP cos(D ) cos(Zt 30 o ) , 30 o D d Zt d 30 o D P . (4.2)
Anytime that #1 is on (including commutation), V1 = 0. The other case of interest to #1 is when

#4 is on (including commutation). For that case, an examination of Figure 4.13 shows that V1 =
-Vdc.
The above analysis is expanded using symmetry and Figure 4.13 to complete the full cycle. The
results are summarized in Table 4.2. Waveforms for several combinations of Į and μ follow the
table.
Page 4-20
June 2006
Table 4.2. Firing Regimes and Corresponding Status of Switches
Angle 1 2 3 4 5 6 Vdc V1 Comment
Case 1. Zt 30 o D • • Vcb Vac #5 on, #6 on

3 #1 turning on,
Case 2. 30 o D d Zt d 30 o D P C C • Vbn 0
2 #5 turning off
Case 3. 30 o D P d Zt d 30 o D 60 o • • Vab 0 #6 on, #1 on

3 #2 turning on,
30 o D 60 o d Zt d 30 o D 60 o P • C C Van 0
2 #6 turning off
30 o D 60 o P d Zt d 30 o D 120 o • • Vac 0 #1 on, #2 on

3 #3 turning on,
30 o D 120 o d Zt d 30 o D 120 o P C • C Vcn 0
2 #1 turning off
30 o D 120 o P d Zt d 30 o D 180 o • • Vbc Vab #2 on, #3 on

3 #4 turning on,
30 o D 180 o d Zt d 30 o D 180 o P C • C Vbn –Vdc
2 #2 turning off
30 o D 180 o P d Zt d 30 o D 240 o • • Vba –Vdc #3 on, #4 on

3 #5 turning on,
30 o D 240 o d Zt d 30 o D 240 o P C • C Van –Vdc
2 #3 turning off
30 o D 240 o P d Zt d 30 o D 300 o • • Vca –Vdc #4 on, #5 on
30 o D 300 o d Zt d 30 o D 300 o P C • C 3 –Vdc #6 turning on,

Vcn #4 turning off
2
Notes. Van is the reference angle. Į = 0 when Vac swings positive (i.e., 30˚). . Firing angle Į can be as large as 180˚. Symbol C
in the above table implies a commutating switch. Symbol • implies a closed switch that is not commutating.
Page 4-21
June 2006
Page 4-22
June 2006
For rectification,
the rise during
commutation is
concave
Page 4-23
June 2006
30 degrees
Page 4-24
June 2006
30 degrees
Page 4-25
June 2006
90 degrees
Page 4-26
June 2006
150 degrees
For inversion,
the rise during
commutation
is convex
Page 4-27
June 2006
Kimbark’s Equations and the Thevenin Equivalent Circuit
As can be seen in the graphs, Vdc has a period of 60˚. Then, the average value, Vdcavg, can be
found by integrating over any period. Using the table and the period
30 o D d Zt 30 o D 60 o ,
§ S S S ·
3¨ 3 T D P T D ¸
vbn (T )dT ³ S6 3 v ab (T ) dT ¸ ,
S ¨¨ 2 ³T
Vdcavg 6
S
(4.3)
D T D P ¸
© 6 6 ¹
§ S S S ·
3 ¨ 3 V LLP T D P 2S T D S ¸
¨ ³ S
6 sin(T )dT V LLP ³ S6 3 sin(T ) dT ¸
S ¨ 2 3 T D 3 T D P 6 ¸
© 6 6 ¹
3V LLP 3 § S 2S S 2S
¨ cos( D P ) cos( D )
S 2 © 6 3 6 3
S S S S S ·
cos( D ) cos( D P ) ¸
6 3 6 6 6 ¹
3V LLP § 3 § S S · § 2S S · ·¸
¨ ¨ cos(D P ) cos(D ) ¸ ¨ cos(D ) cos( D P ) ¸¸
S ¨© 2 © 2 2 ¹ © 3 3 ¹¹
3V LLP § 3 3 2S 2S
¨ sin( D P ) sin(D ) cos(D ) cos( ) sin( D ) sin( )
S ¨© 2 2 3 3
S S ·
cos(D P ) cos( ) sin(D P ) sin( ) ¸
3 3 ¹
3V LLP § 3 3 1 3 1 3 ·
¨ sin( D P ) sin(D ) cos(D ) sin( D ) cos( D P ) sin(D P ) ¸
S ¨© 2 2 2 2 2 2 ¸
¹
3V LLP § 1 1 ·
¨ cos(D P ) cos(D ) ¸ , leaving
S ©2 2 ¹
3V LLP
Vdcavg cos(D ) cos(D P ) (4.4)
2S
Now, for current, evaluating (4.1) at the end of commutation, i.e., Zt 30o D P , yields
i1 (Zt 30 o D P ) I dc
V LLP
2ZL

cos(D ) cos(30 o D P 30 o ) , so that
VLLP
I dc cos(D ) cos(D P ) (4.5)
2ZL
Page 4-28
June 2006
To develop the Thevenin equivalent circuit for the DC side, recognize that at no load, I dc 0,
and if there is no DC current, then ȝ = 00˚.
˚. Thus, from (4.4), the open-circuit (Thevenin
equivalent) voltage is
3V LLLP
3V
VTTH cos((D ) . (4.6)
S
If the Thevenin equivalent circuit exists, then it must obey the Thevenin equation
Vdcavg VTH RTH I dc .
Substituting in for Vdcavg , VTH , and I dc yields
3V LLP 3V LLP V LLP

cos(D P ) cos(D ) cos(D ) RTH cos(D ) cos(D P ) ,
2S S 2ZL
R S R S
cos(D P ) cos(D ) 2 cos(D ) TH cos(D ) TH cos(D P ) .
3ZL 3ZL
Gathering terms,
§ R S· § R S·
cos(D P )¨1 TH ¸ cos(D )¨1 2 TH ¸ 0.
© 3ZL ¹ © 3ZL ¹
Note that if
3ZL
RTH , (4.7)
S
then the above equation is satisfied, leaving the DC-side Thevenin equivalent circuit shown
below.
RTH
T 3V LLLP
VTTH cos((D )
S
I dc + 3ZL
VTTH Vdcavg RTH
S
– V LLLP
33V
Vdcavg
d cos(D ) cos(D P )
2S
V LLP
I dc cos(D ) cos(D P )
2ZL
Page 4-29
June 2006
Power is found by multiplying (4.4) and (4.5), yielding
3V LLP V
Pdcavg Vdcavg I d cos(D ) cos(D P ) x LLP cos(D ) cos(D P ) , so
2S 2ZL
2
Pdc
3VLLP
4SZL

cos 2 (D ) cos 2 (D P ) . (4.8)
Since the converter is assumed to be lossless, then the AC power is the same as (4.10).
To estimate power factor on the AC side, use
3V LLP
P Vdcavg I dc cos(D ) cos(D P ) x I dc
pf true 2S
3Vline neutral rms I rms V V
3 LLP I rms 3 LLP I rms
3 2 2
3 cos(D ) cos(D P ) I dc
x (4.9)
2S I rms
To approximate the rms value of current, it is very helpful to take advantage of the symmetry of
the waveform. The shape of ia (t ) is similar to that shown below. Since it is half-wave
symmetric, only the positive half-cycle need be shown.
I dc
2S
i (Zt ) I dc i (Zt )
3
S S S 2S S 2S
D D P D D P
6 6 6 3 6 3
For small ȝ (i.e., ȝ < 20º), the commutating portions of the current waveform can be
approximated as straight-line segments.
Page 4-30
June 2006
I dc
2S
i (Zt ) I dc i (Zt )
3
S S S 2S S 2S
D D P D D P
6 6 6 3 6 3
2 1
Remembering that the rms value a triangular wedge of current is I avg I 2pp , the rms value of
2
the above waveform becomes
§1 2 1 2 · 2 § 2S · §1 2 1 2 ·
¨ I dc I dc ¸ P I dc ¨ P ¸ ¨ I dc I dc ¸ P
2 © 4 2 ¹ © 3 ¹ © 4 2 ¹
I rms |
S
2 §2 P ·
I dc ¨ ¸.
© 3 2S ¹
Therefore,
2 P
I rms | I dc ,
3 2S
and for ȝ < 20º, then
2
I rms | I dc (4.10)
3
with an maximum error of approximately 5%.
Substituting (4.12) into (4.11) yields
3 cos(D ) cos(D P ) I dc 3 § cos((D ) cos((D P ) ·

pff ttru
uee
true x ¨ ¸. (4.11)
2S 2 S© 2 ¹
I
dc 3
Page 4-31
June 2006
True power factor is the product of distortion power factor pf dist and displacement power factor
pf disp . Thus, examining (4.11), the conclusion is that
3
pf ddist , (4.12)
S
and
cos((D ) cos((D P )
pf ddisp . (4.13)
2
Analysis of Notching
L1 L2
Van ia
a1
L1 L2
ib
Vbn
b1
ic L1 L2
Vcn
c1
Nothing is a phenomenon of interest mainly when sensitive loads are operated near a converter
and share a portion of the converter’s Thevenin equivalent impedance.
Assume that each L of the converter is divided into two inductances, L1 and L2, and that a
sensitive load is located at a1, b1, c1. Thus, L1 represents the fraction of the Thevenin
equivalent impedance that is shared between the converter and the sensitive load. The objective
is to determine the voltage notching present in line-to-neutral voltage Va1n and in line-to-line
voltage Va1b1.
Line-to-Neutral Voltage Notching
From KVL,
di a d (i1 i4 ) di1 di

Va1n Van L1 Van L1 Van L1 L1 4
dt dt dt dt
Current ia is zero or constant, and thus Va1n = Van, except when i1 or i4 are commutating. As
shown previously, these commutation currents and times are
i1
V LLP
2ZL

cos(D ) cos(Zt 30 o ) , 30 o D d Zt d 30 o D P ,
Page 4-32
June 2006
i1
V

I dc LLP cos D cos(Zt 150 o ) , 150 o D d Zt d 150 o D P ,
2ZL
and by symmetry 180˚ later when
i4
V LLP
2ZL

cos(D ) cos(Zt 210 o ) , 210 o D d Zt d 210 o D P ,
i4
V

I dc LLP cos D cos(Zt 330 o ) , 330 o D d Zt d 330 o D P .
2ZL
Thus, when #1 is commutating,
Va1n Van
L1 V LLP
L1 L2 2

sin Zt 30 o
§
V LNP ¨¨ sin Zt
2 L
3L1
L

sin Zt 30 o ·¸¸ for 30o D d Zt d 30o D P , (4.14)
© 1 2 ¹
and similarly,
§ ·
Va1n V LNP ¨¨ sin Zt
3L1
2L1 L2

sin Zt 150 o ¸¸
© ¹
§ ·
V LNP ¨¨ sin Zt
3L1
2L1 L2

sin Zt 30 o ¸¸ for 150 o D d Zt d 150 o D P . (4.15)
© ¹
When #4 is commutating,
§
3L1
2L1 L2

sin Zt 210 o ·¸¸ for 210o D d Zt d 210o D P ,
© ¹
§
2 L
3L1
L

sin Zt 330 o ·¸¸ for 330o D d Zt d 330o D P .
© 1 2 ¹
Rewriting,
§
3L1
2L1 L2

sin Zt 30 o ·¸¸ for 210o D d Zt d 210o D P , (4.16)
© ¹
Page 4-33
June 2006
§
3L1
2L1 L2

sin Zt 30 o ·¸¸ for 330o D d Zt d 330o D P . (4.17)
© ¹
Summarizing
Va1n Van except when
§
3L1
2L1 L2

sin Zt 30 o ·¸¸ ,
© ¹
for 30 D d Zt d 30 D P , and for 210 D d Zt d 210 o D P ,
o o o
§
3L1
2L1 L2

sin Zt 30 o ·¸¸
© ¹
for 150 o D d Zt d 150 o D P , and for 330 o D d Zt d 330 o D P .
Sample graphs forr Va1n are shown on the following pages.
Line-to-Line Voltage Notching
The easiest way to determinee Va1b1 is to recognize thatt Vb1 is identical to Va1 except for being
shifted by 120˚, and to then subtractt Vb1 from
m Va1. The expressions are not derived. Rather,
sample graphs for Va1b1 using graphical subtraction are shown in the following figures.
Page 4-34
June 2006
Line-to-Neutral Voltage Notching LS1/(LS1+LS2) = 1
1.5
alpha = 30
mu = 15
1
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
1.5
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
-2

LS1/(LS1+LS2) = 0.5
1.5 alpha = 30
mu = 15
1
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
1.5
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
-2
Page 4-35
June 2006
LS1/(LS1+LS2) = 1
1.5
alpha = 150
mu = 15
1
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
1.5
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
-2

LS1/(LS1+LS2) = 0.5
1.5 alpha = 150
mu = 15
1
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
1.5
0.5
0
0 30 60 90 120 150 180 210 240 270 300 330 360
-0.5
-1
-1.5
-2
Page 4-36
Grady Chapter 5. Effects and Symptoms Page 5-1
June 2006
5. Effects and Symptoms
5.1. Utility
Harmonics-related problems
p on electric utility distribution systems are usuallyy created byy
primary-metered customers. Typically,
yp y these problems are due to 500kVA (and ( larger)
g ASDs or
induction heaters. In weaker systems, or near the end of long feeders, 100 – 200kVA nonlinear
g to create pproblems. The significant
loads may be sufficiently large g harmonics are almost always
th th th th th
5 , 7 , 11 , or 13 , with the 5 harmonic being the problem in most instances.
y
Classic utility-side y p
symptoms p
of harmonics problems are distorted voltage waveforms, blown
capacitor fuses, and transformer overheating. Capacitors are sensitive to harmonic voltages.
Transformers are sensitive to harmonic currents.
Typical utility-side symptoms are described in the next few pages.
x Resonance
Consider the resonant case shown in Figure 5.1, where the rectangular current injection
of a 5000HP six-pulse current-source ASD produced voltage resonance on a 25kV
distribution system. The 13% THDV caused nuisance tripping of computer-controlled
loads, and the 30% THD I distortion caused overheating of parallel 25kV/480V
3750kVA transformers that supplied the ASD. The dominant voltage harmonics are the
13th (8.3%), and the 11th (7.0%).
Page 5-1
June 2006
13th Harmonic, 8.3%

11th Harmonic, 7.0%
Oscilloscope Image of Six-Pulse LCC Current Waveform

and the Resulting Voltage Resonance
Fourier Series Reconstruction Using Harmonics through the 17th

(Although the phase angles were not recorded, they were manually adjusted in the
plotting program to match the oscilloscope waveshapes.)
Figure 5.1. Resonance Due to 5000HP Six-Pulse Line-Commutated ASD
Page 5-2
June 2006
Resonance occurs when the harmonic currents injected
j by nonlinear loads interact with
system impedance to produce high harmonic voltages. Resonance can cause nuisance
tripping of sensitive electronic loads and high harmonic currents in feeder capacitor
banks. In severe cases, capacitors produce audible noise and sometimes bulge.
To better understand resonance, consider the simple parallel and series cases shown in the
one-line diagrams of Figure 5.2. Parallel resonance occurs when the power system
presents a parallel combination of power system inductance and power factor correction
capacitors at the nonlinear load. The product of harmonic impedance and injection
current produces high harmonic voltages.
Series resonance occurs when the system inductance and capacitors are in series, or
nearly in series, from the converter point of view.
For parallel
p resonance,, the highest
g g distortion is at the nonlinear load. However,
voltage
for series resonance,, the highest
g voltage
g distortion is at a remote ppoint,, pperhapsp miles
awayy or on an adjacent
j feeder served byy the same substation transformer. Actual feeders
can have five or ten shunt capacitors,
p , so manyy parallel
p and series paths
p exist,, making
computer simulations necessary to predict distortion levels throughout the feeder.
High
THDV
Parallel Resonance (high

( g voltage g distortion at converter load,
low voltage distortion at points sown the feeder)
High
THDV
Series Resonance (low

( voltage
g distortion at converter load,
high voltage distortion att points down the feeder)
Figure 5.2. Simple Examples of Parallel and Series Resonance
Page 5-3
June 2006
A simple parallel resonant case would be, for example, where the only capacitors on a
feeder are at the converter location. In such cases, it is possible to use the parallel
resonance approximation formula developed below.
Let Lsys be the total per phase series inductance “seen” at the converter load connection
point. Lsys is determined from the short circuit duty at the bus. If C ccapp is the
capacitance per phase of the power factor correction capacitor, the peak of the parallel
resonant curve occurs at
Z res 1
f res
r .
2S 2S Lsys
s C cap
c
The inductive reactance of the system at the fundamental frequency is
X sys
X sys Z1 Lsys , so Lsys
s .
Z1
The reactance of the capacitor at the fundamental frequency is
1 1
X cap , so C cap
c .
Z1Ccap Z1 X cap
Substituting into the f res equation yields
1 Z1 Z1 X cap X ccap
f rres Z1 X cap f1 .
2S X sys 2S X sys X sys
s
In the per unit system, system short circuit MVA is
1
ASC
MVAS ,
X sys
s
and capacitor MVA is
1
Accap
MVA ,
X ccap
so that the resonant frequency expression becomes
Page 5-4
June 2006
MVASC
S
f rres f1 . (5.1)
MVA
ACAP
C
stiifff systems” (i.e., relatively high MVASC ) have higher resonant frequencies.
Thus, “stiff
When capacitors are added, the resonant frequency is lowered.
The risk of using (5.1) is that it represents only a small part of the true harmonics
situation. Three important points to remember are
1. While ((5.1)) ppredicts a resonant frequency,

q y, it ggives no information about the
broadness of the resonant curve. Thus,, if a system
y is resonant at,, for example,
p the
th
6 harmonic,, one might g innocently y conclude there is no harmonics pproblem.
However,, due to the broadness off the resonance curve, the 5th and 7th harmonics
will be greatly affected.
2. Anytime
y there are shunt capacitors,
p , there are resonant frequencies.
q In fact,, almost
g y resonant near the 5th and 7th harmonics.
all distribution feeders are strongly
However, resonance is a problem only y if there are sufficient harmonic amperes to
excite harmonic voltages so that THD v exceeds 5%.
3. Most utilityy distribution feeders have five or more capacitor

p banks,, so that there
are manyy parallel
p and series ppaths. Thus,, computer
p q
simulations are required to
accuratelyy predict
p g the feeder and adjacent feeders
distortion levels through
connected to the same substation transformer.
To illustrate the broadness of the resonance curve, consider the case shown in Figure 5.3.
This curve represents the Thevenin equivalent impedance, also known as the “driving
point impedance,” at the customer bus. The situation is simple parallel resonance. Note
that as the amount of power factor correction is increased by adding additional kVArs,
the peak of the resonance curve moves toward lower frequencies.
Figure 5.3 illustrates the following two important facts concerning resonance:
1. The resonance curve is very broad.
2. Typical
yp ppower factor correction practices
p to the 0.95-0.98 DPF range will cause
distribution feeders to resonate near the 5th and 7th harmonics.
Page 5-5
June 2006
Customer Bus
Adding Q lowers the resonant

frequency and the impedance
at the 5th harmonic
200%
dZ(5th)
5th
Figure 5.3. Thevenin Equivalent Impedance at Customer Bus
Figure
g 5.3 is also useful in estimating
g the harmonic voltages
g that will exist at the
customer bus. Consider,, for example,
p , the 0.95 power
p factor correction case. At the 5th
harmonic,, the drivingg point
p impedance
p is approximately
pp y 200% (i.e.,
( , 2 pu).
p ) If the
converter load is 0.18 pu, then the 5th harmonic current will be (assuming the 1/k rule)
0.18
0.036 pu. The 5th harmonic voltage estimate is then 0.0366 x 2 0.072 pu. Thus, a
5
th
h
5 harmonic voltage of 7.2% can be expected, meaning that the THD v will be at least
7.2%. Of course, the THD v will be higher after the contributions of the 7th, 11th, and 13th
(and higher) harmonics are included.
Page 5-6
June 2006
x Nuisance Tripping of Sensitive Loads
Some computer-controlled loads are sensitive to voltage distortion. Although it is

difficult to find reliable data on this subject, one case showed that a THDV of 5.5%
p
regularly shut down computerized g ppipe
lathes at a large p company
p y heat treatment
operation.
p While voltageg distortions of 5% are not usually a problem, voltage distortions
above 10% will cause significant nuisance tripping.
x Blown Capacitor Fuses, Failure of Capacitor Cells, and Degredation of Internal

Capacitance
A common harmonics-related complaint comes from capacitor crew foremen or other

distribution feeder maintenance personnel
p who complain
p that “a capacitor
p bank has to be
rebuilt often,”
often, or “fuses on a capacitor bank blow regularly,
regularly,” or “a capacitor bank hums,”
hums,
or “the capacitance of a bank is diminishing.”
Harmonic voltagesg produce

p exaggerated
gg harmonic currents in capacitors
p because of the
inverse relationship between capacitor impedance and frequency. To illustrate this point,
the measured current waveform of a 300kVAr, 480V bank at a commercial bank building
is shown in Figure 5.4. The waveform is dominated by an 11th harmonic (23.3%). The
principal distorting load in the building was a large UPS.
Capacitors with excessive harmonic currents often produce a load humming noise.
Although the human ear is relatively insensitive to 60Hz, it is quite sensitive to the 5th
harmonic and above (i.e., 300 Hz and above).
11peaks
Figure 5.4. 300 kVAr, 480V Capacitor Current Waveform at Commercial Bank Building
Page 5-7
June 2006
1
Since capacitor impedance varies according to , then the impedance for harmonic k
jZC
1
is , where Z1 is the fundamental radian frequency (e.g., 120S radians/sec for 60
jkZ1C
Hz systems). Because of this inverse relationship, moderate harmonic voltages can
produce large currents in capacitors. For example, if a capacitor has 10% voltage
distortion due entirely to the 5th harmonic, the induced 5th harmonic current is 0.10 x 5 =
0.50 pu on the capacitor base. The corresponding rms current in the capacitor increases
to 12 0.50 2 = 1.12 pu times the fundamental current. A 10% 11th harmonic voltage
produces an even greater rms current, 1.49 pu.
Now, consider an example where voltage distortion on a capacitor is assumed to be

divided among six-pulse characteristic harmonics through the 25th, in inverse proportion
to frequency. This assumption implies that the harmonic currents have equal magnitudes.
Since the voltages are expressed in per unit of fundamental, the squared voltage THD is
THDv2 V52 V72 V11

2 2
V13 2
V17 2
V19 2
V23 2
V25 .
m are assumed to vary inversely with

Because the harmonic voltages in this example
frequency, then
5 5
V7 V5 x , V11 V5 x , etc., so
7 11
§ 52 52 52 52 52 52 5 2 ·¸
THDv2 V52 x ¨1 ,
¨ 7 2 112 13 2 17 2 19 2 23 2 25 2 ¸
© ¹
THDv2
THDv2 V52 x 2.108 , so V52 .
2.108
THDv
Taking the square root, V5 , then the current on the capacitor base is
1.452
5 x THDv
I 5 ( pu ) 5V5 ( pu ) .
1.452
Since all eight harmonic currents in this example are equal, the total squared rms
capacitor current, including fundamental, is
2
2 § 5 x THDv ·
I rms ( pu ) 12 8 x ¨ ¸ 1 94.9 x THDv2
© 1.452 ¹
Page 5-8
June 2006
The square root of the above formula is used to computer rms currents for a range of
voltage distortion values, and the results are given in Table 5.1.
Table 5.1. RMS Capacitor

p g the 25th) versus Voltage
Current ((in ppu for harmonics through
Distortion (assuming that voltage harmonics decrease in proportion to frequency)
RMS Capacitor Current

THD v pu
0.00 1.000
0.05 1.112
0.10 1.396
Thus,, it is reasonable to expect

p a 40% increase in capacitor rms current when voltage
distortions are in the 10% range.
Capacitors may also fail because of overvoltage stress on dielectrics. A 10% harmonic
voltage for any harmonic above the 3rdd increases the peak voltage by approximately 10%
because the peak of the harmonic often coincides, or nearly coincides, with the peak of
the fundamental voltage.
x Transformer Overheating That Cannot Be Explained by kVA Load Level Alone
Another common harmonic-related complaintp is that “my transformer is only 70%

loaded, but it is too hot to hold my hand on.” These cases are usually limited to situations
where the transformer serves a large nonlinear load.
There are two reasons for overheating.
1. Losses in a conductor increase when harmonics are present because losses are
proportional to the square (at least) of rms current, and rms current increases with
current distortion according to
I rms THD I2 .
1 T (5.2)
2. Because of the resistive skin effect and winding

g pproximity
y effect,, one ampere
p of
harmonic current produces more losses than does one ampere of fundamental current.
The impact of harmonic currents on transformers is more serious than on conventional

conductors because the resistive skin effect is enhanced within closely-spaced
y p
transformer windings.
g Good engineering
g g practice
p calls for the derating of transformers
that serve nonlinear loads to an equivalent 80% of nameplate kVA.
Page 5-9
June 2006
To examine the rms current effect, consider a transformer that serves an ideal six-pulse
1
converter with classical harmonic magnitude currents. In terms of the fundamental
k
current I1, rms , the squared rms current is
2 § 1 1 1 1 1 1 ·
I rms I12, rms x ¨¨12 ¸¸ . (5.3)
© 5 2 7 2 112 13 2 17 2 19 2 ¹
The above infinite series converges to the form
2
S
2
I rms I12, rms x 1.0966 I12, rms .
9
Since losses increase by the square of rms current, the winding losses automatically
increase to at least 1.0966 times the fundamental-only case. Thus, if losses are to be held
constant at their rated value so that transformer heating is not excessive, the rms current
1
(and equivalent kVA rating) should be lowered to at least = 0.955pu. of
1.0966
nameplate. If harmonics above the 25th harmonic are ignored, the equivalent kVA rating
is 0.960 pu (i.e., practically the same as the infinite series case).
However, the major transformer derate comes from the resistive skin effect. The resistive
skin effect occurs because higher-frequency currents migrate to the outermost portions of
a conductor, increasing its equivalent resistance. For power transformers, this
phenomenon is usually modeled by dividing resistance into two parts – a non-frequency
dependent part, and a frequency-dependent part. The frequency-dependent part is
assumed to increase in proportion to the square of frequency, as given by
Rk R DC x (1 k 2 PEC R ) (5.4)
where Rk is the winding resistance at harmonic k, R DC is the winding resistance at DC,

and PEC R is the winding eddy current loss factor. PE
EC R ranges from 0.01 for low
g service transformers with relatively small conductors to 0.10 for substation
voltage
transformers having large conductors.
Since heating is proportional to squared current times resistance,, the above variation can
be incorporated into an equivalent rms current that takes into account skin effect.
Incorporating (5.4) into (5.3) yields equivalent rms current
§ f
1 (1 k 2 x PEC R ) ·¸
2
I rms , equiv I12, rms x ¨12 ¦ x . (5.5)
¨ k 2 (1 1 x PEC R ) ¸
© k 5,7,11,13, ¹
Page 5-10
June 2006
The above series does not converge if PEC R z 0 . Thus, it is appropriate only to discuss
a finite number of terms, such as through the 25th harmonic. Using the square root of
(5.5) to give the derate, and ignoring harmonics above the 25th, the equivalent kVA rating
for a realistic range of PEC R is given in Table 5.2.
Table 5.2. Equ

Equivalent kVA Rating for Transformers Serving Six-Pulse Loads
(ignoring harmonics above the 25th, and assuming that current
harmonics decrease in proportion to frequency)
Equivalent kVA Rating

PEC R pu
0.00 0.960
0.02 0.928
0.04 0.900
0.06 0.875
0.08 0.853
0.10 0.833
5.2. End-User
Symptoms experienced by end-users include the utility symptoms described above, plus the
items described on the next few pages.
x Digital Clocks Gaining Time
Digital clocks work off the principle of counting zero crossings or slope changes in the
60Hz fundamental voltage. There may be some filtering present in the clock circuitry,
but if voltage harmonics are strong enough, then it is possible to have multiple zero
crossings or slope changes that cause the clocks to run fast. Older digital clocks are
reported to be the most sensitive to voltage harmonics.
An example of a voltage waveform that was responsible for this phenomenon is shown in
Figure 5.5. The waveform has a 2%, 36th harmonic.
Page 5-11
June 2006
Figure 5.5. Voltage Waveform That Caused Digital Clocks to Gain Time
x Telephone Interference
Telephone interference has been a harmonics-related concern for many decades, but the
gradual phasing out of open-wire telephone circuits has reduced the number of
interference problems. While the frequency response of the combined telephone circuit
and human ear is largely immune to 60 Hz interference, higher harmonics fall into the
low-audio range.
When harmonic currents on power lines inductivelyy couple p into nearbyy phone
p lines, they
can cause significant
g interference. Typically,
yp y, the problem
p harmonics are either
characteristic six-pulse
p g converters, or 9th and higher multiples
harmonics due to large p of
( , zero sequence)
three (i.e., q ) due to transformer saturation. All things
g being
g equal,
q , zero
q
sequence harmonics are more pproblematic than ppositive and negative
g q
sequence harmonics
because a-b-c zero sequence fields are additive and, therefore, do not decay as rapidly
with distance.
The telephone influence factor (i.e., TIF) curve shown in Figure 5.6 gives the relative
interference weighting that applies to inductively-coupled harmonic currents flowing in
power lines.
Page 5-12
June 2006
12000
10000
8000
TIF
6000
4000
40th 50th
2000
harmonic harmonic
0
0 600 1200 1800 2400 3000 3600 4200
Fre que ncy - Hz
Figure 5.6. Telephone Influence Factor (TIF) Curve
Telephone interference problems are usually solved by the telephone company, in

cooperation with the electric utility involved. Solutions are often trial-and-error and
usually consist of moving or disconnecting capacitor banks that have high harmonic
currents, or placing tunable reactors in the ground path of wye-connected capacitor banks
that have high harmonic currents. The tuning reactors are invisible to positive and
negative sequences, but they can change the zero-sequence resonant frequency of a
distribution feeder and often eliminate the resonance problem.
x Motor Heating
For frequencies higher than fundamental, three-phase induction motors can be

approximated by positive/negative shunt impedances,
Zk Rwinding jkX " ,
where Rwinding is the motor winding resistance, and X " is the fundamental frequency
subtransient reactance (typically 0.20 pu on motor base). Since most motors are three-
wire delta or ungrounded-wye connected, motors appear as open circuits to zero sequence
harmonics.
Assuming X " = 0.20, relatively small Rwinding with respect to kX " , and a 5th harmonic
voltage of 10%, the induced 5th harmonic current will be
0.10
I 5, rms 0.10pu on the motor base.
5 x 0.20
Page 5-13
June 2006
Thus, harmonic voltages can create additional rotor winding currents and increase the
I 2 Rwinding losses in three-phase motors by several percent.
High
g efficiencyy single-phase
g p induction motors are more sensitive to voltage harmonics
than are three-phase motors. Auxiliary parallel windings with series-run capacitors create
a quasi-sinusoidal flux wave to improve efficiency. The series auxiliary winding
inductance and run capacitor create a series resonant path in the 4th – 11th harmonic
range.
To illustrate this phenomenon, current waveforms for a 2HP, 230V, fully loaded motor
were measured, with and without significant 5th harmonic voltage applied. Waveforms
for both cases are shown in Figure 5.7. The strong 5th harmonic current causes additional
heating and produces noticeable audible noise.
20
15
10
5
Amperes
-5
-10
-15
-20
Applied 5th Harmonic Voltage = 1.3%. Resulting THD I = 11.6%.
20
15
10
5
Amperes
-5
-10
-15
-20
Applied 5th Harmonic Voltage = 5.4%. Resulting THD I = 34.0%.
Figure 5.7. Sensitivity

y of Fully
y Loaded 2HP, 230V Single-Phase High-
Efficiency Induction Motor Current to 5th Harmonic Voltage
Page 5-14
June 2006
x Overloaded Neutral Conductors in Commercial Buildings
In a three-phase, four-wire system, the sum of the three phase currents returns through the
neutral conductor. Positive and negative sequence components sum to zero at the neutral
point, but zero sequence components are purely additive in the neutral.
Power system engineers are accustomed to the traditional rule that “balanced three-phase
systems have no neutral currents.” However, this rule is not true when zero sequence
harmonics (i.e., primarily the 3rd harmonic) are present. In commercial buildings with
large numbers of PC loads, the rms neutral current can actually exceed rms phase
currents.
Consider the measurements shown in Figure 5.8 for a 120/208V service panel feeding g
147 PC workstations plusp some miscellaneous linear load. The rms currents for pphases a,
b,, and c are 40.2A,, 52.7A,, and 47.8A, respectively. The neutral current is mainly 3rdd
harmonic and has rms value 60.9A, which is approximately 1.3 times the average
fundamental a-b-c phase current.
Page 5-15
June 2006
150 150
100 100
50 50
0 0
-50 -50
-100 -100
Phase C Current - Amperes
Phase A Current - Amperes

THD I = 633.0%
-150
(courtesy of Mr. Russ Ehrlich, Connectiv Solutions) -150
Phase A at Service Panel, 40.2Arms Fundamental. THD I = 77.4%. Phase C at Service Panel, 47.8Arms Fundamental. THD I = 60.7%.
150
150
100
100
50
50
0 0
Amperes
-50 -50
-100 -100
Phase B Current - Amperes

-150 -150
Phase B at Service Panel, 52.7Arms Fundamental. THD I = 54.7%. Sum of Phases A,B,C in Neutral Wire (bold curve) at Service Panel, 60.9
Arms. THD I = 633.0%.
Figure 5.8. Current Measurements for a 120/208V Service Panel Feeding 147 PC Workstations
Page 5-16
June 2006
Large numbers off PC loads cause a flattening of phase-to-neutral voltages inside

commercial buildings, as shown in Figure 5.9. The spectrum is given in the Appendix.
Flattened
200 voltage actually
reduces current
150
distortion
100
Volts 50
-50
-100
-150
-200
Figure 5.9. Phase A Voltage at Service Panel for 147 PC Workstations
This commonly-observed flattened voltage waveform has THDV 5.1% and contains
3.9% of 5th, 2.2% of 3rd, and 1.4% of 7th harmonics.
RMS neutral current due to 3rd harmonics can be higher than that of the example shown
in Figure 5.8. Many PCs have 3rd harmonic currents greater than 80%. In these cases,
the neutral current will be at least 3 • 80% = 240% of the fundamental a-b-c phase
current.
Thus, when PC loads dominate a building circuit, it is good engineering practice for each
phase to have its own neutral wire, or for the common neutral wire to have at least twice
the current rating of each phase wire.
Overloaded neutral currents are usuallyy onlyy a local problem

p inside a building,
g, for
example
p on a service panel.
p At the service entrance,, the harmonic currents pproduced by
PCs and other nonlinear loads are diluted byy the many linear loads including air
conditioners, pumps, fans, and incandescant lights. The current waveform @ 480V for a
large office and classroom building is shown in Figure 5.10. The THD I is 7.0%,
consisting mainly of 4.5% of 5th, 3.7% of 7th, and 3.5% of 3rd harmonics. The spectrum
is given in the Appendix.
Page 5-17
June 2006
1500
About 600-700kVA three-
phase. Even though the
1000
building contains many
PCs and other nonllinear
500
loads, dilution with linear
Amperes
loads reduces net THDI
0
to 7.0%
-500
-1000
-1500
Figure 5.10. Phase A Current at 480V Service Entrance to College of Business

Administration Building, University of Texas at Austin
Page 5-18
Grady Chapter 6. Conducting an Investigation Page 6-1
June 2006
6. Conducting an Investigation
Electric utility engineers may be confronted with harmonic problems on their own distribution
feeders, or within customer facilities. Distribution feeder cases are the most difficult to deal with
since a large harmonics source can pollute the voltage waveform for many miles, including
adjacent feeders connected to the same substation transformer. Customer facility cases are the
simplest to investigate because the distances are smaller and the offending load can usually be
identified by turning candidates off-and-on while observing area voltage waveforms with an
oscilloscope or spectrum analyzer.
The focus of this section is on investigating distribution feeder problems where the combination
of harmonic current injection and resonant networks act together to create objectionable
harmonic levels.
6.1 Field Measurements
In some cases, field measurements alone can be used to identify the source of a harmonics
problem. To do this, consider the following:
p
1. It is important to remember that utility-side
y harmonics pproblems are almost always
y
created by primary-metered customers, and the culprit is usually a 500kVA (or higher)
ASD, rectifier, or induction furnace. Therefore, if the problem appears suddently, it is
prudent to ask questions within your company to find out which large customers on the
feeder (or adjacent feeders) may have added a large distorting load.
2. It is wise to make field measurements before contacting the customer. The basic tool
needed is a portable spectrum analyzer that can monitor and record harmonic voltages
and currents. Voltage measurements can be made at capacitor control circuits or at
metering points. The frequencies of interest (e.g. 1500Hz and below) are low enough
g, control,, and service transformers accurately
that standard metering, y pportray
y feeder
g waveforms. Usually,
voltage y, the feeder voltage
g distortion is high
g near the harmonics
source,, but when there are manyy shunt capacitors,
p , remote ppoints mayy also have high
g
voltage
g distortion. It is desirable,, but perhaps impossible, to turn off all shunt capacitors
when the measurements are made.
3. Next, it is prudent to monitor and record voltage and current harmonics at the customer’s
metering point for at least two days, and perhaps more. These data will help to correlate
the customer’s daily work shift patterns or nonlinear loads with distortion levels. THDV ,
THD I , the 5th and 7th harmonic magnitudes, and if possible, harmonic power, should be
recorded. The main indicator is the customer’s THD I .
4. While there is debate on the subject, most power quality engineers believe that harmonic
power is a good indicator of the source of harmonics. In fact, if a distribution feeder has
one large distorting load, then that load is the source of all harmonic power on the feeder.
Some spectrum analyzers compute harmonic power. If the customer is the source of
harmonic power, then you can expect the net harmonic power (a few percent of
Page 6-1
June 2006
fundamental power) to flow out from the customer onto the feeder, further comfirming
that the harmonics source is inside the customer’s facility. The 5th harmonic usually has
the largest harmonic power.
If Steps 1 – 4 are inconclusive,, then it is sometimes possible

p to “track down” a harmonics source
byy takingg harmonic power
p measurements at convenient points alongg the feeder.
f For example,
voltage measurements can be made at capacitor control boxes. Current measurements can be
p
made with fiber optic-linked current transformers that connect directlyy to the feeder conductors.
Using g voltage
g and current,, net harmonic power
p can be calculated. The expectation is that the net
harmonic power flows away from the source.
A wire loop (i.e. “search coil”) can be connected to the voltage input of a spectrum analyzer to
dM
monitor the current-induced N signal that exists below an overhead feeder. The search coil
dt
has been used for decades byy telephone
p companies
p to detect the ppresence of high
g harmonic
currents. While the search coil ggives no ppower orr voltage
g information,, it is useful because large
harmonic currents exist on either or both sides of a resonatingg capacitor
p bank. Resonating
p
capacitor banks are sometimes turned off, moved, or filtered in an attempt to relieve the
harmonics problem.
6.2. Computer Simulations
Field measurements are useful when a harmonics problem alreadyy exists. However,, computer p
simulations are needed to studyy ppotential pproblems in advance. For example,
p , if a customer
desires to add a 1000HP ASD,, then an advance studyt is definitely needed so that problems can
be resolved before the ASD is installed. A harmonics study proceeds in much the same way as a
loadflow, short circuit, or motor starting study.
Unless a distribution system is badly unbalanced,, or there is a veryy large

g single-phase
g p harmonics-
p
producingg load such as an electric train, harmonics analysis can usually be performed using the
balanced assumption. The reasons are that
x Most problem-causing loads are large three-phase balanced loads such as ASDs.
x Distribution capacitors are usually applied in the form of three-phase banks, having a
balancing effect on harmonics propagation.
x Phase identification of single-phase loads and load levels may not be available or easily
obtained.
x The quality of harmonics data for the distorting loads may be poor, so that injection
“rules of thumb” must be used.
x Systems are often studied in advance, so that not all of the actual data are available.
Page 6-2
June 2006
In spite of these difficulties, experience has shown that distribution feeder harmonic simulations
match “real world” measurements veryy well,, and that simulation is a reliable tool for studying
y g
solutions such as ppassive filtering.
g The term “accurate” ggenerallyy means that simulated voltage
distortions match field measurements within a few percent (on a 100% base).
To obtain this accuracy, these eight rules must be obeyed:
1. When modelingg a distribution feeder,, include in your

y studyy all the feeders attached to the
same substation transformer,, and in equal
q detail. On the transmission side of the
substation transformer, establish a simple Thevenin equivalent using the short circuit
impedance. Transmission line capacitance can be added on the substation high-side, but
it usually is not important to the study results.
2. Ten to twenty aggregated busses per feeder is usually adequate detail.
3. Load distributions along actual feeders are not known with great accuracy. However,,
total feeder kVA load and kVA ratings of individual transformers are known. Load
distributions are often estimatedd byy assumingg that the total feeder kVA load is distributed
in proportion to individual load transformer ratings. Adjustments should be made for
large customers whose daily load profile does not track the feeder load profile.
4. Harmonics models for conventional loads must be included. These can be simple shunt
resistances, where the resistances are sized according to active power.
5. The worst-case for harmonics is usually y when the harmonics-producing loads are at full
power, and the conventional loads are at low power. Conventional loads add damping
and reduce distortion levels, and their sinusoidal currents dilute the nonlinear load
currents.
m
6. Capacitor banks are very important and must be included in the study. Usually this
means a case with all capacitors on, and a case with only the fixed capacitors on. Other
likely capacitor scenarios may also be needed.
g off underground
7. If there are significant lengths g cables,, cable capacitances may be
important and should be lumped onto the trunk feedersf p
in the form of shunt capacitors.
p
“Important” is relative to the size of the other shunt capacitors. 100kVAr is a good rule
for being “important.” The capacitance of power cables can be estimated using Table
6.1.
Page 6-3
June 2006
Table 6.1. Capacitance and Charging of 12.47kV and 25kV Cables
Cable Capacitance kVAr Capacitance kVAr
(12.47kV) (12.47kV) (25kV) (25kV)
1/0 0.163 9.56 0.124 29.2
4/0 0.222 13.02 0.162 38.2
350 kcmil 0.267 15.65 0.1914 45.1
500 kcmil 0.304 17.82 0.215 50.7
1000 kcmil 0.401 23.5 0.278 65.5
Capacitance: μF per km per phase
kVAr: (three-phase) per km
8. When the system has multiple sources operating

p g at various power
p levels ((i.e.,, 10 or more
), then it is important to consider harmonics cancellation brought about by phase
sources),
angle diversity. The simplest way to consider diversity is to multiply the harmonic
injection currents for each load by the following:
x 3rd harmonic, multiply by 1.0 (i.e., no diversity)

x 5th and 7th harmonics, multiply by 0.9
x 11th and 13th harmonics, multiply by 0.6
x Higher harmonics, multiply by 0.2
There are two basic techniques for performing harmonics studies – frequency-domain, and time-
domain.
x Frequency-domain modeling is most often used for harmonics studies where the focus of
attention is on the network. Approximate models are used for nonlinear loads. Each
harmonic is studied individually, and the results are superimposed to produce time-
domain waveforms.
x Time-domain modeling is usually performed with full three-phase detail and precise
models of nonlinear loads. Time-domain modeling is often used to study small networks
where the focus of attention is inside specific equipment such as ASDs.
Five-Bus Computer Simulation Example
An industrial customer will be served by constructing a three-mile 12.5kV overhead feeder from
a dedicated 138/12.5kV substation transformer. The customer will have 5MW @ dpf = 0.85 of
conventional load and a 2000HP, six-pulse adjustable-speed drive (ASD). The ASD is
connected through a delta-delta transformer (i.e., no phase shift). The customer also has
1800kVAr of shunt power factor correction capacitors.
The 138kV substation bus has the following characteristics:
x Z+ = 0.4 + j2.5% (100MVA base)
Page 6-4
June 2006
x 50 miles of 138kV transmission lines are connected to it (line charging = 0.0808MVAr
per km).
The dedicated substation transformer has the following characteristics:
x Pbase = 15MVA
x 138kV delta / 12.5kV grounded-wye connection
x 0.95 per unit tap on the 138kV side
x Z+ = 0.5 + j10.5% (on 15MVA base).
The overhead feeder will be constructed with 477 ACSR arm-type construction that has the
following characteristics:
x R+ = 0.1316 : per km
x X+ = 0.387 : per km (@ 60 Hz)
x C+ = 0.01106 μF per km.
The conventional load transformer is rated at 7.5MVA and has Z+ = 0.50 + j5.0% (on 7.5MVA
base).
Once the data have been gathered, the next step is to draw a one-line diagram with all
impedances and loads expressed on a common base. The base values are selected as 10MVA
throughout, and 12.5kV on the feeder section. The voltage base varies throughout the circuit
according to nominal transformer turns ratios.
The swing bus is effectively grounded for harmonics with a j0.01% "harmonics-only subtransient
impedance." The purpose of this grounding impedance is to model the ability of the "far-distant"
system to absorb harmonic currents without incurring appreciable voltage distortion.
Calculations for the above steps are shown below.
Page 6-5
June 2006
For the transmission system,
Swing Bus Short Circuit Z Substation
0.4 % j2.5 %
@ 100 MVA, 138 kV
3.25 MVAR
3.25 MVAR
Swin g Bus 0.04 % j0.25 % Substation
2
(1 3 8)
Z BASE =
10
= 1904.4 :
j0.01% 32.5 % 32.5 %
For the distribution feeder,
R+ = 0.1316 : per km • 1.609 km/mile • 3 miles = 0.635
X+ = 0.387 : per km • 1.609 km/mile • 3 miles = 1.868
C+ = 0.01106 μF per km • 1.609 km/mile • 3 miles = 0.0534μF
ZBASE = (12.5) 2 / (10) = 15.625
§ 12500 ·§ 12500 ·
Line Charging = 3 ¨¨ ¸¸¨¨ ¸¸ 2S (60) 0.054 x 6 VAr
© ¹© ¹
= 3.15kVAr = 0.0315 % @ 10MVA
0.635
Rpu = x 100% = 4.06%
15.625
1.868
Xpu = x 100% = 11.96%
15.625
Page 6-6
June 2006
4.06 % j11.96 %
0.0158 % 0.0158 %
For the substation transformer,
138 / 12.5
Tap = 0.95 pu j 7.0 %
0.33 %
+ Tap = 0.95 pu
Z = 0.5 + j 10.5 %
@ 15 MVA
For the conventional load transformer
12.5 / 4160
0.67 % j 6.67%
Z+ = 0.5 + j 5.0 %
@ 7.5 MVA
The 1800kVAr of shunt power factor correction capacitors becomes 18% on a 10MVA base.
The final one-line diagram is shown in Figure 6.1.
Page 6-7
June 2006
Conventional
Converter 5
Swing Bus Sub HV Sub LV 4 P + jQ =
0.67 + j 6.67 % 50 + j 31.0
V = 102 0 1 2 3 4.06 + j 11.96 %
0.04 + j 0.25 % 0.33 + j 7.0 %

E = 0.0315 %
E = 65 %
P, dpf = 14.92 %, 0.85
j 0.01 % tap = 0.95

( Harmonic
Only Subtransient )
íj18% VAr
j18% VAr
Figure 6.1. System One-Line Diagram for Five-Bus Example

(Data in PCFLOH files *_FIVE.csv)
6.3. Passive Filters
Filters accomplish two objectives – power factor correction, and shunting one or more harmonic
currents to ground. A series tuned filter can be constructed in each phase to ground by placing a
choke in series with a shunt capacitor, and then tuning the choke so that the inductive and
capacitive
p reactances are equal
q but opposite
pp at the desired harmonic. Tuning g a filter slightly
g y
below the desired harmonic,, for example p at the 4.7th instead of the 5th harmonic, helps to reduce
capacitor voltage without significantly degrading ffilter performance. Often the addition of a 4.7th
(i.e., 5th ) filter is sufficient to solve harmonics problems.
g kVAr to the filter. In most cases,, the filter kVAr should be

Care must taken to dedicate enough
approximately the amount needed to power factor correct the nonlinear load. Filters with smaller
kVAr will have sharp tuning curves and will be easily overloaded by stray harmonics that are
present in the network.
p
Since a filter capacitor usuallyy experiences
p p rms voltage, plus significant harmonics,
1.2 to 1.3 pu
care must be taken that the capacitor voltage rating is adequate. The fact that kVArs decrease by
the square of voltage must also be taken into consideration.
To illustrate filter design, the five-bus system is modified by converting the 1800kVAr capacitor
bank into a 4.7th harmonic filter. First, a new bus (#6) is created, and the 1800kVAr capacitor
bank is moved from Bus 4 to Bus 6. In reality, the 1800kVAr bank would be replaced with a
higher-voltage rated bank, with sufficient kVArs so that it produces 1800kVAr at system
voltage. Then, Bus 4 is connected to Bus 6 with a series choke that has the appropriate
reactance.
Page 6-8
June 2006
Conventional
Converter 5
Swing Bus Sub HV Sub LV 4 P + jQ =
0.67 + j 6.67 % 50 + j 31.0
V = 102 0 1 2 3 4.06 + j 11.96 %
0.04 + j 0.25 % 0.33 + j 7.0 %

E = 0.0315 %
E = 65 %
P, pf = 14.92 %, 0.85
j 0.01 % tap = 0.95 6 0.502 + j 25.1 %

( Harmonic
Only Subtransient )
íj18% VAr j18% 4.7th Filter
Figure 6.2. System One-Line Diagram for Five-Bus Example with Filter
(Data in PCFLOH files *_FIVE_FILTER.csv)
The tuning formulas for harmonic k are
X C ( pu @ 60 Hz )
Let kX L ( pu @ 60 Hz ) , so that
k
X C ( pu @ 60 Hz )
X L ( pu @ 60 Hz ) .
k2
In this example,
1
X C ( pu @ 60 Hz ) 5.55 pu , so that
0.18 pu
5.55 pu
X L ( pu @ 60 Hz ) 0.251 pu , or 25.1%.
4.7 2
On a 12.5kV, 10MVA base, the choke inductance (each phase of wye connection) is
1 0.251 x 15.625
L X L x Z BASE x 10.4mH .
2Sf 2S x 60
Assuming
X L ( pu @ 60 Hz )
50
R
Page 6-9
June 2006
25.1%
for the choke, the choke resistance is estimated to be R 0.502% , or 0.0784ȍ. The
50
modified one-line diagram is shown in Figure 6.2.
Filter performance is checked using three steps.
1. Impedance scans are performed, without and with the filter. The filter notch should be at
the design harmonic.
2. The converter bus voltage waveform, without and with the filter, is examined. 5th
harmonic filtering is usually adequate. However, if the voltage distortion is still more
than 4 or 5%, it may be necessary to add a larger 5th filter, or possibly 7th, 11th, and 13th
filters, in that order. Usually, the higher the harmonic, the fewer kVArs committed to a
filter. A good rule for dedicatingg kVAr when multiplep filters are needed is to stairstepp the
kVAr as follows: if Q kVArs are used for the 5th harmonic, then Q/2 should be used for
the 7th, Q/4 for the 11th, and Q/4 for the 13th. Of course, actual sizes must match standard
sizes. The total kVAr should ppower factor correct the nonlinear load. For best
performance, a filter should have at least 300 kVAr (three-phase).
3. The filter current waveform is checked to make sure that it is absorbing the appropriate
harmonic and that the filter current is within rating.
Simulation results for the five-bus system, without and with the filter, are given in Figures 6.3 –
6.5.
Page 6-10
June 2006
":Case Title = ":Case Title =

File Nam e, Bus Num bers and Nam es, File Nam e, Bus Num bers and Nam es,
c:\pcfloh\zbus1_new .csv, #4 CONVERTER , #4 CONVERTER c:\pcfloh\zbus1_new .csv, #4 CONVERTER , #4 CONVERTER
3.0 3.0
2. But a new resonant peak is
created at a lower frequency
2.0 2.0
3. Thus, when filtering, you cannot
in general skip harmonics, e.g.
filter the 7th, but not the 5th
1.0 1.0
Log10(ZMAG-%)
Log10(ZMAG-%)
1. Notch correctly located
at desired frequency
0.0 0.0
1 3 5 7 9 11 13 15 17 19 21 23 25 1 3 5 7 9 11 13 15 17 19 21 23 25

File Nam e, Bus Num bers and Nam es, File Nam e, Bus Num bers and Nam es,
c:\pcfloh\zbus1_new .csv, #4 CONVERTER , #4 CONVERTER c:\pcfloh\zbus1_new .csv, #4 CONVERTER , #4 CONVERTER
180.0 180.0
135.0 135.0
90.0 90.0
45.0 45.0
0.0 0.0
Z Angle
Z Angle
-45.0 -45.0
-90.0 -90.0
-135.0 -135.0
-180.0 -180.0
1 3 5 7 9 11 13 15 17 19 21 23 25 1 3 5 7 9 11 13 15 17 19 21 23 25
Figure 6.3. Impedance Scans at Converter Bus (Without Filter on Left, With Filter on Right),
(plots produced by program PCFLO_ZBUS_PLOT.XLS)
Page 6-11
June 2006

c:\pcfloh\vsoln_new .csv, #4 CONVERTER . Vrm s = 99.3% , c:\pcfloh\isoln_new .csv, #4 CONVERTER , #6 FILTER_4.7 ,
THDV = 7.3% Ckt#1 . Irm s = 19.0% , THDI = 18.4%
Voltage
Branch Current
":Case Title = Figure 6.5. Filter Current Waveform)
c:\pcfloh\vsoln_new .csv, #4 CONVERTER . Vrm s = 99.3% , (plot produced by program
THDV = 3.3%
PCFLO_ISOLN_PLOT.XLS)
Now,
1. Use Excel to examine the PCFLOH input files for this case,
i.e., BDAT_FIVE.csv and LDAT_FIVE.csv.
2. Call up the *_FIVE.csv case in PCFLOH, examine the input summary.

Solve, and then view the voltage and current spectra and waveforms.
Voltage
3. Close PCFLOH, and view some of the output files in Excel.
4. Display voltage and current waveforms with PCFLO_VSOLN_PLOT.xls

and PCFLO_ISOLN_PLOT.xls.
5. Perform an impedance scan on the *_FIVE.csv case. View the results in

PCFLOH, and also in PCFLO_ZBUS_PLOT.xls.
Figure 6.4. Converter Bus Voltage Waveforms Bus
(Without Filter at Top, With Filter at Bottom), 6. Use Excel to view filter files *_FIVE_FILTER.csv. Run PCFLOH to check
(plots produced by program for tuning. View the output with PCFLO_ZBUS_PLOT.xls. Run the full
harmonics study, viewing results in PCFLOH and also in
PCFLO_VSOLN_PLOT.XLS) PCFLO_VSOLN_PLOT.xls and PCFLO_ISOLN_PLOT.xls.
Page 6-12
Grady Chapter 7. Standards and Solutions Page 7-1
June 2006
7. Standards and Solutions
7.1. IEEE 519
q
The most often quoted harmonics standard in the U.S. is IEEE 519,, “Recommended Practices
and Requirements for Harmonic Control in Electric Power Systems.” IEEE 519 attempts to
establish reasonable harmonic goals for electrical systems that contain nonlinear loads. The
objective is to propose steady-state harmonic limits that are considered reasonable by both
electric utilities and their customers. The underlying philosophy is that
x customers should limit harmonic currents,

x electric utilities should limit harmonic voltages,
g
x both parties share the responsibility for holding harmonic levels in check.
IEEE 519 applies to all voltage levels, including 120V single-phase residential service. While it
does not specifically state the highest-order harmonic to limit, the generally accepted range of
application is through the 50th harmonic. Direct current, which is not a harmonic, is also
addressed and is prohibited. Since no differentiation is made between single-phase and three-
phase systems, the recommended limits apply to both.
It is important to remember that IEEE 519 is a recommended practice and not an actual standard
or legal document. Rather, it is intended to provide a reasonable framework within which
engineeers can address and control harmonic problems. It has been adopted by many electric
utilities and by several state public utility commissions.
Definitions and Terms
THD. Total Harmonic Distortion (or Distortion Factor) of voltage or current is the ratio of the
rms value of harmonics above fundamental, divided by the rms value of the fundamental.
PCC. Point of Common Coupling is a point of metering, or any point as long as both the utility
and the customer can either access the point for direct measurements of the harmonic
indices meaningful to both, or estimate the harmonic indices at the point of interference
through mutually agreeable methods. Within an industrial load, the point of common
coupling is the point between the nonlinear load and other loads.
There is some flexibility in determining the PCC, but in most instances, it is at the meter.
An electric utility might also interpret the PCC to be on the high-voltage side of the
service transformer, which would have the effect of allowing a customer to inject higher
harmonic currents.
ISC. Maximum short circuit current at the PCC.
IL. Maximum demand load current (fundamental frequency component) at the PCC,
calculated as the average current of the maximum demands for each of the preceeding
twelve months. For new customers, this value must be estimated.
Page 7-1
June 2006
TDD. Total demand distortion, which is the THD of current (using a 15 or 30 minute averaging
measurement period) normalized to the maximum demand load current IL.
Utility Limits
Electric utilities are responsible for maintaining voltage harmonics and THDV . The limits are
divided into two categories: voltages 69kV and below, and voltages above 69kV. For electric
utility distribution systems (i.e., corresponding to 69kV and below), the limits are
For Voltages 69kV and Below

Individual Total Harmonic
Voltage Distortion
Harmonic THDV
% %
3.0 5.0
Customer Limits
Customers are responsible for maintaining current harmonics and THD I . Again, the limits are
divided into two categories: voltages 69kV and below, and voltages above 69kV. For 69kV and
below, the limits are
For PCC Voltages 69kV and Below

Maximum THD I in % of IL for Odd Harmonics k
ISC/IL k 11 11 d k 17 17 d k 23 23 d k 35 35 d k TDD
20 * 4.0 2.0 1.5 0.6 0.3 5.0
Dedicated
customer
20 - 50 7.0 3.5 2.5 1.0 0.5 8.0
transformer 50 - 100 10.0 4.5 4.0 1.5 0.7 12.0
100 - 1000 12.0 5.5 5.0 2.0 1.0 15.0
t 1000 15.0 7.0 6.0 2.5 1.4 20.0
* All power generation equipment is limited to these values of THD I , regardless of the actual
ISC/IL.
Even-ordered harmonics are limited to 25% of the odd harmonic limits ggiven in the tables.
Loads that produce direct current offset, e.g. half-wave converters, are not allowed.
7.2. Public Utilitity Commission Standards
Several states, including Texas and Oklahoma, have adopted harmonic standards. These
standards are based upon IEEE 519. Texas state ruling 25.51, “Power Quality,” permits an
electric utility to charge a fee for having to investigate and remedy a customer-created excessive
Page 7-2
June 2006
harmonics condition. The fee is limited to actual cost incurred plus a reasonable administrative
cost.
7.3. Interacting with Customers
It is wise for an electric utility to develop a written document of harmonics policy that can be
distributed to large industrial customers as the need arises. While a good basis for the document
is IEEE 519, other procedural items should also be addressed. The following key points should
be considered for inclusion in the document.
Modeling
Data are needed to determine whether a proposed customer’s facility will cause harmonic limits
to be exceeded. These data include
x One-line drawings of the customer’s facilities, showing ratings and connections of all
electrical equipment,
x Location, connection, size, and control method of capacitors,
x Conductor sizes and impedances,
x Location and type
yp of nonlinear loads,
x Overall plant
p load and portion
p that is nonlinear,
x Location, rating, connection, and impedance of transformers.
Customers should be responsible for modeling their systems to project harmonic levels and
determine whether the utility’s harmonic limits will be exceeded.
The utility should provide information regarding the local power system to support the
customer’s modeling efforts. This information should include
x Available fault duty at customer’s location,

x Ultimate available fault current,
x Impedance and ratings of service transformers,
x Possible voltage range variation.
Filter modeling should include the utility’s background voltage distortion allowed by IEEE 519,
which is 3% for a single harmonic and 5% THDV . Failure to include this allowed background
distortion may result in inadequate filter designs.
The utility may need copies of the customer’s harmonic analysis for review prior to approving
the customer’s proposed facilities. The utility may need the customer to submit manufacturer’s
documentation and test data demonstrating the harmonic content of nonlinear loads.
Page 7-3
June 2006
Measurements
The utility should reserve the right to measure the amount of a customer’s harmonic current
injection at any time at the point of common coupling (normally the electric meter). These
measurements are usually spot checks, but additional monitoring may be required.
Mitigation Devices and Methods
The customer should be responsible for the design, installation, operation, and maintenance of
mitigation devices required to meet the utility’s harmonic limits. Mitigation devices may include
current limiting reactors, passive filters, active filters, or other devices that minimize the flow of
harmonic currents onto the utility’s distribution system.
The customer should submit mitigation device maintenance records to the utility upon request.
The installation and testing of mitigation equipment should be subject to the approval of the
utility. The mitigation devices must be capable of handling the IEEE 519 permitted background
voltage distortion that can exist on the utility’s distribution system.
The utility will likely reconfigure the distribution system regularly in response to load changes
and to resolve outages. The mitigation equipment should operate independently of these
changes.
7.4. Solutions
Solution techniques fall into two broad categories – preventive and remedial.
Preventive Measures
Preventive measures focus on minimizing the harmonic currents that are injected into power
systems. Preventive measures include
x Strict Adherence to IEEE 519.
x Phase Cancellation. The use of twelve-pulse converters instead of six-pulse converters

should be encouraged. Most utility harmonic problems are associated with high 5th and
7th harmonic currents, and if they are eliminated through phase cancellation, harmonic
problems rarely develop. In situations where there are multiple six-pulse converters,
serving half of them (in terms of power) through delta-delta or wye-wye transformers,
and the other half through delta-wye or wye-delta transformers, achieves net twelve-pulse
operation.
x Encouragement of Low Distorting Loads. Because of IEEE 519, increasing attention is

being given to the THD I of distorting loads. A customer often has a distortion choice in
loads. For example,, twelve-pulse (or higher) ASDs and low-distortion fluorescent lamp
ballasts can be purchased.
Page 7-4
June 2006
x Computer Simulations. It is always better to simulate the impact of a large distorting
load before it is ordered and installed. Solutions can be proposed and evaluated “on
paper” and perhaps implemented when the load is installed. Once the distorting load is
connected, the customer will likely be under considerable pressure to operate it and
perhaps less likely to commit additional funds to deal with a distortion problem.
Remedial Measures
Remedial measures include
x Circuit Detuning. By using only field measurements such as capacitor current

y the capacitor
waveforms and search coil readings, it is possible to identify p banks that are
p
most affected by resonance. As a temporary y measure to “buyy time” before a real solution
can be found, the affected capacitor bank can be switched off to see if the resonance
problem subsides. Of course, the problem may simply transfer to another capacitor bank,
so post-switching measurements at other capacitor banks must be made to see if the
temporary solution is satisfactory. If switching a capacitor bank off temporarily solves
the problem, computer simulations may be in order to test filtering options and possible
re-location of the capacitor bank.
x Passive Filters. These are widely used to control harmonics, especially the 5th and 7th
harmonics. Most filters consist of series L and C components that provide a single-tuned
notch with a low-impedance ground path. At 50/60Hz, these filters are, for all practical
purposes, capacitors. Thus, passive filters provide both power factor correction and
voltage distortion control.
5th harmonic filtering g is usuallyy adequate

q to solve most distribution system
y harmonic
p
problems. However,, in some cases it mayy be necessaryy to add 7th, 11th, and 13th harmonic
filters,, in that order. In ggeneral,, harmonics mayy not be “skipped.”
pp p , if the
For example,
th th th
p
problem harmonic is the 7 , both 5 and 7 harmonic filters must be added because the
7th filter alone would aggravate the 5th harmonic voltage. Filters tuned near the 3rdd
harmonic must be avoided because transformers and machines located throughout
distribution feeders are sources of third harmonics, and their currents will easily
overwhelm 3rd harmonic filters.
Usually, the higher the harmonic, the fewer kVArs needed for a filter. For multiple p filter
installations,, a good
g p
practice is to stairstepp the kVAr as follows: if Q kVArs are used for
the 5th harmonic,, then QQ/2 should be used for the 7th, QQ/4 for the 11th, and Q/8
Q for the
th
13 . Of course,, actual sizes must match standard kVAr sizes. For best performance, a
filter should be at least 300 kVAr (three-phase).
It may be possible to add low-voltage filters within the confines of an industrial customer
without performing computer simulations, as long as all shunt capacitors in the facility
are filtered. However, in a utility distribution system, it is always prudent to perform
computer simulations to make sure that a filter does not aggravate the harmonics situation
at a remote point. This is especially true if the feeder also has unfiltered capacitors.
Page 7-5
June 2006
Some problems associated with passive filters are that
x their effectiveness diminishes over time as their capacitors age, losing μF and thus
raising their notch frequency,
x they attract harmonic currents from all sources in the network – new, known, and
unknown, so that they may become overloaded.
x Active Filters. This is a new and promising technology, but there are as yet few
distribution feeder installations. Active filters are power electronic converters that inject
equal-but-opposite distortion to yield more sinusoidal voltage waveforms throughout a
network. Active filters have the advantages of
x time-domain operation so that they automatically “tune” to the problem harmonic

or harmonics,
x current limiting capability to prevent overload by new or unknown sources of

harmonics on the network,
x multi-point voltage monitoring so that they can simultaneously minimize

distortion at local and remote busses.
The performance of mitigation equipment must be verified by extensive monitoring, both before
and after commissioning. At least two days of recordings before commissioning, and one week
after, should be made to assure that the mitigation equipment is performing as planned. One
week of measurements is needed so that the entire weekly load cycle can be observed.
Monitoring should include time traces of voltage and current THD, spectra, sample waveforms,
power, and harmonic power.
Economic Justification of Mitigating Measures
From a customer’s perspective, the most common economic justification of harmonics mitigation
is in minimizing down-time due to nuisance tripping of senstive loads. This cost is totally
customer-dependent.
From a loss perspective, harmonics can be considered as a reduction in power factor. In Chapter
3, true power factor was shown to be
dpff1
pff ttrue | .
D I2
1 THD
Thus, the true power factor off nonlinear loads is limited by THD
D I . Consider a nonlinear load
with perfect displacement power factor ( dpf1 ). When current distortion is included, the true
power factor degrades, as shown in Table 7.1.
Page 7-6
June 2006
Table 7.1. Maximum True Power Factor for Nonlinear Loads
THD I Maximum
% pf true
10 0.99
20 0.98
30 0.96
50 0.89
100 0.71
Since the true power factors given above are for the special case of unity dpf1 , they represent
maximum true power factors for nonlinear loads. “Actual” true power factor is the product of
maximum true power factor and displacement power factor, and the product can be significantly
lower than dpf1 .
The power
p factor comparison
p presents
p a ratherr optimistic
p ppicture,, because harmonic currents
actually cause more losses per ampere than do fundamental currents.
Voltage harmonics have been shown to cause additional losses in motors, especially high-
efficiency single-phase motors. Voltage harmonics induce harmonic currents that increase motor
p
losses and insulation temperature. Research byy Dr. Ewald Fuchs at the Universityy of Colorado at
g distortions in the 6% range
Boulder has shown that voltage g with predominant
p 3 and 5th
rd
harmonics can reduce the expected lifetime of single-phase motors by 25%-30%.
Page 7-7
Grady Chapter 8a. System Matrices and Simulation Procedures Page 8a-1
June 2006
8a. System Matrices and Simulation Procedures
If a system has more than a few busses, then simulations are too tedious to perform by hand.
Thus, a computer-based simulation procedure is needed, and the procedure is usually matrix
oriented. This chapter describes the simulation procedure used in PCFLO (and PCFLOH).
8.1. Summary of the Procedure Used in PCFLO
y in per
We perform harmonic analysis p unit on a system
y base. We begin with a conventional
loadflow. The loadflow is an excellent debugging tool – if the loadflow does not solve or if the
answers seem unreasonable, then the input data are questionable. Once solved, the loadflow
g magnitudes
establishes the fundamental voltage g and phase angles through the system. The
“swing bus” is assumed to have no distortion,, and its fundamental frequency
q y voltage
g phase
p angle
y zero. The swingg bus voltage
is always g angle
g serves as the pphase angle reference for all
fundamental and harmonic voltages and currents in the system.
After a successful loadflow, we consider harmonics above the fundamental,, one byy one,, using
q
the standard frequency-domain approach – that is, the principle off current injection
n in
in the
frequency domain, one harmonic at a time. For each harmonic k
x we build the admittance matrix, using the appropriate positive, negative, or zero-sequence
network for harmonic k,
x we use the spectrum and power level of each nonlinear load to determine its kth harmonic
injection current,
x we simultaneously inject the harmonic k currents for all nonlinear loads into the network,
x we determine bus voltages and branch currents for harmonic k using standard phasor
analysis techniques.
After bus voltages and branch currents are determined for all harmonics, the results are
combined using Fourier series’ to produce time-domain waveforms.
To illustrate the procedure, begin by considering Figure 8a.1. Bus m is a “linear bus” with a
conventional P,Q linear load represented by impedance Zm,linear
, . For harmonics purposes, in
PCFLO we model the linear P loadd as a resistor to provide harmonics damping, and we ignore
the linear Q load. Bus n is a “nonlinear bus” that has a harmonic current injector. The harmonic
current injector creates harmonic voltages throughout the power system. Our objective is to
determine those harmonic voltages.
There may be many linear and nonlinear busses in a system, and busses can contain both linear
loads and nonlinear loads. Mixed busses are easily handled in the solution by simply placing
harmonic current injectors in parallel with Zm,linear impedance terms.
Page 8a-1
June 2006
Linear bus m Power system with Nonlinear bus n

many linear and
+ nonlinear busses +
Zm,linear Vm,k ⁄įm,k Vn,k ⁄įn,k In,k ⁄șn,k
í í
x For linear loads, Zm,linear is a resistive load model for the linear P load, in pu at
harmonic k.
x At a generator, Zm,linear is the subtransient impedance, in pu, of the generator.
x At the loadflow swing bus, Zm,linear is a very small number, i.e., j0.0001pu, so
that the voltage at the swing bus remains a perfect, distortion-free sinewave.
x Zm,linear may be a parallel combination of loads and subtransient impedances.
Figure 8a.1. Illustrative System for Harmonic k, k > 1
8a.2. Determining the Magnitudes and Phase Angles of the Harmonic Injection Currents
Once the loadflow is solved, then the local fundamental voltage magnitude and phase angle at
bus n, Vn,1G n,1 , is known. Nonlinear loads are specifiedd in terms of P, dpf, and spectrum.
Define d n,1 as the lagging dpf angle of the nonlinear load. Then the fundamental current
magnitude of the nonlinear load in per unit is
Sn Pn
I n,1 (8a.1)
Vn,1 Vn,1 x dpf n
and the phase angle of the nonlinear load fundamental current with respect to the swing bus is
G n,1 d n,1 . (8a.2)
Now, we consider the spectrum for the nonlinear load current at harmonic k,
I spec, k : magnitude of harmonic k current, in per unit of fundamental load current,

T spec, k : angle of harmonic k current, with respect to fundamental load current.
The magnitude of the load current for harmonic k is then
Page 8a-2
June 2006
Pn
I nn,, k I nn,,1 x I spec
s c,, k x I spec
s c,, k . (8a.3)
Vnn,,1 x ddpff n
Time shifting to account for the shift in fundamental phase angle yields the phase angle of the
harmonic load current,
T n, k T spec, k k x G n,1 d n,1 . (8a.4)
Finally, because the harmonic current is typically modeled as an injector (as shown in Figure
8a.1) instead of a load, the current magnitude must be negated by adding (or subtracting) 180º to
the phase angle, yielding the final expression for phase angle,
T n,k
,k , k k x G n,1 d n,1 r 180q .
T spec,k (8a.5)
8a.3. Determining Network Voltages
Network phasor voltages for each harmonic are determined by using Kirchhoff’s Current Law
(KCL) as follows. Consider the three-bus, five branch, one current-source network shown in
Figure 8a.2. The objective is to solve for the three bus voltages with respect to ground.
ZA ZC
1 2 3
ZE ZB ZD
I3
Figure 8a.2. Three-Bus Network
For every harmonic, we apply KCL at the three nodes. The equations are
V1 V1 V2
At bus 1, 0,
ZE ZA
V2 V2 V1 V2 V3
At bus 2, 0 ,
ZB ZA ZC
V3 V3 V2
At bus 3, I3 .
ZD ZC
Collecting terms and writing the equations in matrix form yields
Page 8a-3
June 2006
ª 1 1 1 º
« 0 »
«ZE Z A ZA » ªV1 º ª0º
« 1 1

1

1

1 »« »
V2 «0» , (8a.6)
« ZA Z A Z B ZC ZC »« » « »
« « » «¬ I 3 »¼
« 1 1 1 »» ¬V3 ¼
0
«¬ ZC Z C Z D »¼
or in abbreviated form,
YV I , (8a.7)
where Y is the admittance matrix,, V is a vector of phasor bus voltages (with respect to ground),
g ),
and I is a vector of phasor current injections. Equation (8a.7) must be written for every harmonic
in the study, and the impedances and current injections vary accordingly.
Voltage sources, if present, can be converted to current sources using the usual Thevenin/Norton
conversion rules. If a bus has a zero-impedance voltage source attached to it, then the bus
voltage is already known, and the dimension of the problem is reduced by one.
A simple observation of the structure of the above admittance matrix leads to the following rule
for building Y:
1. The diagonal terms of Y contain the sum of all branch admittances connected directly to
the corresponding bus.
2. The off-diagonal elements of Y contain the negative sum of all branch admittances
connected directly between the corresponding busses.
These rules make Y very simple to build using a computer program. For example,
p , assume that
the impedance data for the above network has the following form, one data input line per branch:
From To Branch Impedance (Entered

Bus Bus as Complex Numbers)
1 0 ZE
1 2 ZA
2 0 ZB
2 3 ZC
3 0 ZD
g FORTRAN instructions would automatically build Y

The following Y, without the need of
manually writing the KCL equations.
Page 8a-4
June 2006
COMPLEX Y(3,3),ZB,YB
DATA Y/9 * 0.0/
1 READ(1,*,END=2) NF,NT,ZB
YB = 1.0 / ZB
C MODIFY THE DIAGONAL TERMS
IF(NF .NE. 0) Y(NF,NF) = Y(NF,NF) + YB
IF(NT .NE. 0) Y(NT,NT) = Y(NT,NT) + YB
IF(NF .NE. 0 .AND. NT .NE. 0) THEN
C MODIFY THE OFF-DIAGONAL TERMS
Y(NF,NT) = Y(NF,NT) - YB
Y(NT,NF) = Y(NT,NF) - YB
ENDIF
GO TO 1
2 STOP
END
Of course, error checking is needed in an actual computer program to detect data errors and bus
numbers. Also, if bus numbers are not compressed (i.e. bus 1 through bus N), then additional
logic is needed to internally compress the busses, maintaining separate internal and external (i.e.,
user) bus numbers.
Usingg harmonic current injections

j from (8a.3)
( ), the kth harmonic bus voltage
) and (8a.5),
( g pphasors
for the entire network are found simultaneously using Gaussian elimination and backward
substitution on (8a.7), or by using direct inversion with
V ZI , where Z Y 1 . (8a.8)
Once the bus voltages are known, then branch currents can be easily calculated.
To assist in understanding how Y is built, Y for the five-bus example in Figure 6.1 is given in
Table 8.1. For simplicity of presentation here, phase shift due to the substation transformer
connection is not included in the table. In reality, PCFLO handles transformer
f phase shift with a
complex transformer tap. Phase shift can also be taken into account by a simple bookkeeping
procedure where the phasor voltages are advanced or delayed accordingly, after a solution is
obtained.
Also, for simplicity of presentation here, the branch resistances in the table are shown to be
p
independent of harmonic k. In reality, one of the data inputs to PCFLO is for the user to specify
the harmonic for which a resistance doubles – with that information, PCFLO varies branch
resistance as a function of k .
Page 8a-5
June 2006
Table 8a.1. Admittance Matrix for the Five-Bus Example in Figure 6.1 (in per unit) for Harmonic k
1 0.65
+ jk
jk 0.0001 2 1
1 0 0 0
1 0.0004 jk 0.0025
+
0.0004 jk 0.0025
1
0.0004 jk 0.0025
1 1 1 0 0
2 +
0.0004 jk 0.0025 0.0033 jk 0.07 0.0033 jk 0.07
0.65
+ jk
2
1
0.0033 jk 0.07
0 1 1 1 0
3 0.0033 jk 0.07 + 0.0406 jk 0.1196
0.0406 jk 0.1196
0.000315
+ jk
2
1
0.0406 jk 0.1196
0 0 1 1 1
4 0.0406 jk 0.1196 + 0.0067 jk 0.0667
0.0067 jk 0.0667
0.000315
+ jk + jk 0.18
2
1
0 0 0 1 0.0067 jk 0.0667
5
0.0067 jk 0.0667 + 0.50
Page 8a-6
June 2006
8a.4. Physical Significance of the Impedance Matrix
Impedance
p matrix Z,, evaluated for the fundamental frequency,
q y, is the key
y to short circuit studies.
Individual Z matrices for each harmonic are important
p to harmonic studies. For harmonic
studies, the diagonal elements of Z are the Thevenin-equivalent impedances
impedances at the system busses,
as we will see below.
Individual elements of Z describe how the voltage at a bus is related to the current injection at
that bus or at other busses. To see this, examine the mth row of V ZI , which is
N
Vm ¦ z m, n I n z m,1 I1 z m,2 I 2 z m,3 I 3 z m, N I N . (8a.9)
n 1
If all current sources are “off” except at bus n, then
Vm z m, n I n , where I j 0 for j 1,2, , N , j z n . (8a.10)
The situation is illustrated in Figure 8.3, where I n, k is a current source attached to bus n, Vm, k
is the resulting voltage at bus m, and all other harmonic current sources are turned off. When
m n , z n, n is by definition the Thevenin equivalent impedance at bus n for harmonic k. When
an “impedance
p scan” is requested
q for a bus ((typically
yp y a nonlinear bus),
), PCFLO computes
p the
q
Thevenin equivalent p
impedance p of 0.2 of a harmonic (i.e., integer and
for that bus in steps
noninteger harmonics 1.0, 1.2, 1.4, … , 24.6, 24.8, 25.0).
Linear bus m Nonlinear bus n

Power system with
many busses
+ +
Zm,linear Vm,k ⁄įm,k Vn,k ⁄įn,k In,k ⁄șn,k
í í
Figure 8a.3. Illustrative System for Harmonic k, k > 1,

with All Sources Off Except at Bus n
For interconnected networks (as is the case for positive and negative-sequences)
negative-sequences),)), current
j
injection y bus in the network. For that reason, Z tends
g at every
at one bus affects the voltage is to be
a full matrix for positive and negative-sequences. On the other hand, zero-sequence networks
q
are often not interconnected. For example, zero-sequence g and currents cannot pass
voltages p
through a transformer with a delta winding. Thus, zero-sequence Z matrices are often sparse.
Page 8a-7
June 2006
8a.5. Creating Time-Domain Waveforms from the Phasor Solution
Once the harmonic phasors are known for a voltage or current, the corresponding time-domain
waveform is easily created. Consider bus m with phasor voltages
Harmonic Phasor Voltage

1 m,,1 ⁄⁄įįm,
Vm,1 m,1
,1
5 m,,5 ⁄⁄įįm,
Vm,5 m,5
,5
7 m,,7 ⁄⁄įįm,
Vm,7 m,7
,7
… …
If the spectra and phasor references are sine series’, then the time domain waveform is
vm (t ) Vm,1 sin((Z1t G m,1 ) Vm,5 sin(5Z1t G m,5 ) Vm,7 sin(7Z1t G m,7 ) (8a.11)
8.6. Comments on Zero-Sequence Networks
Zero-sequence
q networks are affected byy grounding.
g g Ungrounded
g shunt elements,, loads,, and
capacitors
p are “invisible” to zero sequence
q and are not included in the admittance matrix. Delta
connections are open circuits to zero-sequence.
Transformer connections are very important in zero-sequence networks. Figure 8a.4 is shown
below as a reminder of the zero-sequence models for transformers.
Page 8a-8
June 2006
Grounded Wye - Grounded Wye R + jX
Grounded Wye - Delta R + jX
Grounded Wye - Ungrounded Wye

R + jX
Ungrounded Wye - Delta R + jX
Delta - Delta R + jX
Figure 8a.4. Zero Sequence Impedance Equivalent Circuits for Three-Phase Transformers
Page 8a-9
Grady Chapter 8. Case Studies Page 8-1
June 2006
8. Case Studies
Four actual case studies are presented. They are
Case 1. 5000HP Chiller Motor ASD on 25kV System. This case has severe voltage resonance
with nuisance tripping of sensitive loads and overheating of the ASD transformers. Simulations
compare favorably with field measurements.
Case 2. 2000HP Oil Pipeline Pumping Station ASD on 12.5kV System. There are no serious
harmonics problems. Simulations compare favorably with field measurements.
Case 3. Television Broadcast Station on 25kV System. There is a serious problem where the
broadcast picture wobbles due to interaction of a constant-voltage transformer with a distribution
system that was upgraded from 12.5kV to 25kV. Simulations support a quickly-implemented
field solution.
Case 4. 12.5kV Ski Area with 5150HP of Distributed Lift Motor ASDs. This is a planning
study where simulations are used to design a harmonics mitigation strategy. This case illustrates
the effectiveness of phase cancellation and passive filters.
Case 1. 5000HP Chiller Motor ASD on 25kV System
Overview. This is a serious resonance case where a 3.9MW six-pulse line commutated ASD
driving a chiller motor excites resonance in the 25kV underground distribution system shown in
Figure 8.1. The ASD is served by feeder 2203, but the resonance problem is intense throughout
the entire subsystem served by S.W. Substation Transformer #1 (i.e., feeders 2202, 2203, and
2204). The subsystem load at the time harmonics measurements are taken is 9.5MW @ dpf =
0.855, consisting off 5.6MW of conventional load plus the ASD that draws 3.9MW @ dpf =
0.830. The subsystem contains 16.27 miles of three-phase underground trunk and lateral cables.
The voltage and current waveforms at the ASD were shown previously in Figure 5.1 of Chapter
5, and the voltage distortion throughout the 25kV subsystem is approximately 10%. Voltage
g at full ppower and the conventional load
distortion is highest at night, when the chiller is running
( , harmonics damping
(i.e., p g load)) is at low power.
p g voltage
The high g distortion regularly trips off a
computer-controlled automated train system in the late night hours. Also, there is concern that
the resonant overvoltages will cause cable dielectric failure.
The chiller is centrally located so that it can be easily switched from feeder 2203 to feeder 2606,
which is served by the N.W. substation. The switching arrangement is shown in Figure 8.2.
Unfortunately, field switching of the ASD from the preferred feeder (2203) to the secondary
feeder (2606) did not improve the harmonics situation and simply transferred the problem from
one subsystem to the other.
Page 8-1
June 2006
Figure 8.1. 25kV System Serving 5000HP Chiller Motor ASD
Page 8-2
June 2006
Figure 8.2. Local Connections for 5000HP Chiller Motor ASD
Page 8-3
June 2006
Two parallel
p 3750kVA transformers,, totalingg 7500kVA,, serve the ASD. Even though the
5000HP motor draws less than 5000kVA,, the parallel
p transformers overheat. Thus, Case 1
contains the following three classic symptoms of a harmonics problem:
1. Resonance.
2. Nuisance tripping of sensitive loads.
3. Overheating of transformers that cannot be explained by kVA load alone.
By making voltage distortion measurements at the substation while the ASD is turned off and on,
electric utility engineers confirm that the ASD is the source of the harmonics problem.
The unusual feature of this case is that there is a resonance problem

p even though
g no power
p factor
correction capacitors
p are installed. Capacitors
p are not neededd because the underground
g cables
provide considerable power factor correction, especially during low-load periods.
To examine the impact of cables, consider the perfect coaxial case, where each meter has
capacitance
2SH o H r 2S x 8.854 x 2.25

C | = 125pF/meter/phase (or 0.125μF/km/phase),
§ ro · 1
ln¨¨ ¸¸
© ri ¹
and where H o is the permittivity of free space, H r is the relative permittivity of the cable
dielectric, and ri , ro are the coaxial inner and outer radii of each phase. For three phases, the
corresponding kVA at 25kV is
2
§ 25000 ·
Q3I 2
3V LN ZC3 x ¨¨ ¸¸ x 120S x 125 x 10 12 29.5 VAr (three-phase)/meter, or
© 3 ¹
29.5kVAr (three phase)/km.
The net three-phase cable charging of the 16.27 miles of three-phase cables is then 0.772MVAr.
However, a review of the cable manufacturer’s data for this particular case gives the following
values per meter for the three types of 25kV cables:
Table 8.1. Manufacturer-Supplied Electrical Characteristics of 25kV Underground Cables

25kV Cable R+ X+ C kVAr
1/0 Al 0.696 0.1581 0.161 37.9
4/0 Al 0.352 0.1447 0.202 47.6
1000 kcmil Al 0.0827 0.1148 0.354 83.4
R+, X+: Positive sequence resistance and inductive reactance, ȍ per km

C: Capacitance, μF per km per phase
kVAr: Cable kVAr (three-phase) per km
Page 8-4
June 2006
Thus, the actual cable capacitance, which takes into account actual geometries, is considerably
higher than that of the ideal coaxial formula.
It is important to note that the above range of C, 0.161 – 0.354μF/km/phase, is very large
compared to the approximately 0.01μF/km/phase for overhead distribution feeders. The
capacitance of overhead distribution feeders can usually be ignored in harmonics studies,
especially if the feeders have power factor correction capacitors.
The subsystem contains 6.37 miles of 1000 kcmil Al,, 5.40 miles of 1/0 Al, and 4.50 miles of 4/0
Al cables. Thus, the total cable charging is 1.53MVAr. Based on the substation P and dpf , the
P and dpf of the converter, and the 1.53MVAr of cable charging, the dpf of the conventional
load is estimated to be 0.770 (ignoring reactive power losses).
Simulations. (PCFLOH Files *_DFW.csv). Both the S.W. and N.W. systems are measured and
simulated, and the comparisons for both systems match quite well. For brevity, only the results
for the S.W. system are described here.
p
A study of the feeder blueprints j load and circuit branch busses, plus the
shows there are 29 major
ASD bus. This set off 30 busses becomes the “retained” load busses for the study. Loads and
cable charging for non-retained busses are lumped onto the nearest retained load bus. Retained
g
busses are connected with line segments havingg the pper meter characteristics shown in Table 8.1.
The 5.60MW,, 4.67MVAr of conventional load is distributed over the retained load busses
(excluding the ASD) in proportion to net load transformer rating.
The ASD bus is connected to the feeder by the parallel 3750kVA transformers. Each
transformer has impedance 0.79 + j5.69% on its own base, and has connection type high-side
grounded-wye, low-side delta. The impedance of S.W. Transformer#1 is 1.803 + j40.8% on a
100MVA base, and the connection is high-side delta, low-side grounded-wye. The short circuit
impedance on the 138kV side of S.W. Transformer#1 is 0.277 +j1.588% on a 100MVA base.
The 138kV bus is assumed to have voltage 1.025pu. One-half of the line charging of the 138kV
lines connected to the substation bus is 4.97MVAr.
Other nonlinear loads in the subsystem

y are ignored.
g For harmonics purposes,
p p conventional loads
are modeled as resistive elements, sized according to their active power. The ASD injection
current waveform employed is that shown in Section 5, Figure 5.1.
An impedance scan (i.e., Thevenin equivalent impedance) at the 25kV converter bus, shown in
Figure 8.3 as “No KVAr added,” predicts strong parallel resonance at the 12th harmonic.
Page 8-5
June 2006
1800 kVAR No kVAR

Added Added
2.5
Log10 (Impedance - %) 2
1.5
0.5
0
1 3 5 7 9 11 13 15 17 19 21 23 25
Harmonic
Figure 8.3. Thevenin Equivalent Impedance vs. Frequency at 25kV
Converter Bus, with No kVAr Added, and with 1800kVAr Added
The corresponding simulation results show voltage distortions in the 25kV subsystem to be in the
narrow range of 9.1 – 9.7%. Simulated voltage distortion at the 4160V ASD bus, behind the
additional impedance of the parallel 25kV/4160V transformers, is 14.9%.
A comparison with measurements at the S.W. #1 25kV substation bus is given in Tables 8.2 and
8.3. Simulations and field measurements match reasonably well.
Table 8.2. Measured and Simulated Harmonic Voltages at S.W. #1 25kV Substation Bus
No kVAr No kVAr 1800 kVAr 1800 kVAr
Added. Added. Added. Added.
Measured V Simulated V Measured V Simulated V
Harmonic % % % %
5 2.6 3.1 3.2 4.0
7 1.7 2.1 3.0 4.6
11 5.8 6.1 3.1 2.4
13 7.2 5.3 1.3 1.2
17 2.0 2.1 0.6 0.7
THDV 9.9 9.1 5.6 6.7
Page 8-6
June 2006
Table 8.3. Measured and Simulated Harmonic Currents Through S.W. #1 25kV Substation
Transformer
No kVAr No kVAr 1800 kVAr 1800 kVAr
Added. Added. Added. Added.
Measured I Simulated I Measured I Simulated I
Harmonic % % % %
5 9.3 12.7 13.0 18.5
7 5.0 6.2 9.7 15.0
11 12.0 11.4 5.9 4.9
13 11.0 8.4 2.0 2.2
17 2.6 2.5 0.8 0.9
THD I 19.6 20.1 17.4 24.4
In an attempt to reduce the resonance problem by detuning the feeder, 1800kVAr of capacitors
are switched on at the converter 25kV bus. The simulated resonant curve shifts, as shown in
Figure 8.3, and the predicted voltage distortions across the 25kV system lower somewhat to the
6.7-7.6% range. Individual harmonics at the S.W. #1 25kV substation bus are shown in Tables
8.2 and 8.3. Unfortunately, detuning provides only a partial solution because the 5th and 7th
harmonic injection currents are strong, and moving the resonant curve too close to them will only
make the situation worse.
The major uncertainties in this modeling effort are
1. cumulative effect of other,, smaller nonlinear loads that are not included in the study,
2. conventional load level, distribution, and model (for harmonics purposes),
3. cable capacitance.
Page 8-7
June 2006
Case 2. 2000HP Oil Pipeline Pumping Station ASD on 12.5kV System
Overview. A 2000HP six-pulse line-commutated ASD is connected to Bus 5 on the 12.5kV

distribution system shown in Figure 8.4. The three-phase trunk portions of the two feeders
connected to the substation transformer are entirely overhead construction and consist of
x 1.87 miles of 477 ACSR armless construction,

x 2.02 miles of 4/0 ACSR armless construction,
x 4.90 miles of 4/0 ACSR arm construction,
x 0.57 miles of 1/0 ACSR arm construction.
Harmonic measurements are taken on a hot summer day. The transformer load is 10.08MW,
5.47MVAr (i.e., dpf = 0.879), which includes 1.5MW ASD with assumed dpf = 0.83 (i.e.,
1.05MVAr). Subtracting the ASD load from the transformer load leaves 8.58MW, 4.42MVAr
for the conventional load plus capacitors. The uncorrected power factor of the conventional load
is estimated to be 0.800, or 6.44MVAr. Thus, there are likely about 6.44 – 4.42 = 2.02MVAr of
shunt capacitors (plus enough to overcome line and transformer reactive power losses) in
operation when the measurements are taken.
The two feeders have nine shunt capacitor banks, totaling 7.95MVAr. 5.10MVAr of the
capacitors are time-controlled or time-temperature
p controlled,, and 2.85MVAr are regulated
g by
y
either voltage,
g , current,, or power
p factor. At the time of the measurements,, time-controlled and
time-temperature
p controlled capacitors
p are supposed
pp to be on, and the regulated capacitors are
most likely off due to good voltage and power factor levels. For purposes of comparing
simulations to measurements, it is assumed that all of the 5.10MVAr of time and time-
temperature capacitors are on-line. In relation to Figure 8.4, these are
x 600kVAr at Bus 3
x 600kVAr at Bus 4
x 600kVAr at Bus 7
x 600kVAr at Bus 20
x 1200kVAr at Bus 21
x 600kVAr at Bus 24
x 900kVAr at Bus 25
Simulations. (PCFLOH Files *_WGAR.csv). The connected load transformer kVA

information is not readily available. As an approximation, it is assumed that the 8.58MW,
4.42MVAr of conventional load is uniformly distributed over all 12.5kV busses except the
substation transformer buss. The conventional loads are modeled for harmonic purposes as shunt
resistors. Single-phase laterals, underground cable segments from distribution poles to service
transformers, and overhead line capacitances are ignored. The ASD is modeled as a 1/k-rule
harmonics injector.
Page 8-8
June 2006
2000HP Six-Pulse
ASD Here
Highest THDV
points
Figure 8.4. 12.5kV System Serving 2000HP Pipeline ASD
Page 8-9
June 2006
A 10MVA base is chosen. The positive/negative sequence Thevenin equivalent of the 138kV
system (excluding the substation transformer) is 0.05 + j0.344%, with voltage 1.04pu. A shunt
capacitance equal to one-half of the combined line charging of the 138kV transmission lines
connected to the transformer (i.e., 24% on 10MVA base) is placed as a shunt element on the
transformer 138kV bus. The substation transformer has positive/negative sequence impedance
0.312 + j6.75%.
Voltage distortion measurements are taken at the substation 12.5kV transformer bus and at the
ASD. A comparison of measurements and simulations is given in Table 8.4
Table 8.4. Measured and Simulated Voltage Distortion Levels in Oil Pipeline Distribution
Feeders
Measured at Simulated at Measured at Simulated at
ASD ASD Substation Substation
Bus Bus Transformer Transformer
THDV - % THDV - % THDV - % THDV - %
3.3 3.6 2.1 1.9
The highest simulated THDV for this case is 4.0% at Busses 8 and 26. It is interesting to note
that Bus 26 is the most distant bus from the converter and is on an adjacent feeder served by the
same substation tranformer.
Two other cases are simulated. When all capacitors are on, the highest THDV is 4.3% at Bus 7.
When only the regulated capacitors are on, the highest THDV is 5.7% at the ASD bus. Hence,
when a feeder has switched shunt capacitors,
x the highest voltage distortion may be at a remote point from the harmonics source,
x the highest distortion case may be a situation when only a subset of the capacitors are on-
line.
Since the voltage distortions are not objectionable, no additional work is needed in this case.
Page 8-10
June 2006
Case 3. Television Broadcast Station on 25kV System
Overview. The policy of this electric utility is to avoid 25kV delta, grounded-wye transformers
because of potential ferroresonance problems. When a metropolitian area overhead distribution
system is upgraded from 12.5kVto 25kV, all delta, grounded-wye service transformers are
replaced with grounded-wye, grounded-wye transformers.
In this case,, the voltage

g upgrade
pg and transformer replacement
p coincides with the appearance of
an annoying flicker on the broadcast signal of a television staion.
station. The problem is traced to
fluctuations on the output of a 480V/480V saturable reactor transformer that is supposed to
maintain constant 480V ((± 1%)) voltage g to the transmitter rectifier. Measurements show that the
neutral current on the pprimary y of the saturable transformer rises from 30A before the upgrade
pg to
rd
325A after the upgrade,
pg , and that the neutral current is pprimarilyy 3 harmonic. Furthermore, the
primary
p y phase
p current rises to 500A,, while the load current remains 380A. The saturable
transformer hums loudly after the upgrade.
Utility engineers suspect that the problem is harmonics-related. Several capacitor banks on the
feeder are switched off,, and the situation improves
p but does not disappear.
pp In an attempt
p to solve
the problem
p quickly,
q y, a special-ordered
p 25kV/480V delta,, ggrounded-wye
y service transformer is
installed in place of the new grounded-wye, grounded-wye transformer, and the problem
disappears. Thus, it is speculated that the problem is due to harmonics, and specifically to the
zero-sequence 3rd harmonic.
Simulations. (PCFLOH files *_CHIL.csv). At this point, simulations are performed to confirm
that the “quick fix” transformer replacement can be explained. While there is no easy way to
determine how the control system of the saturable transformer reacts with 3rd harmonic voltages,
g
it is relatively easy to show with simulations that the Thevenin impedance of the 480V bus is
x not affected by the 25kV upgrade, but

x greatly affected by the replacement of the service transformer.
With the delta,, grounded-wye

g y transformer,, the impedance
p “seen” byy the saturable transformer is
p y the grounded
simply g impedance
p of the 25kV/480V service transformer. However,, with the
y , grounded-wye
grounded-wye,
g g y service transformer, the impedance is the service transformer
impedance plus system impedance.
The study system has three feeders served by a substation transformer. The total load is
17.8MW, 11.0MVAr, dpf = 0.85. There are 15 power factor correction capacitor banks,
totaling 9.3MVAr. The Thevenin equivalent impedance on the 138kV side of the substation
transformer is 0.0376 + j0.2547% for positive/negative sequence, and 0.094 + j0.6368% for zero
sequence, on a 10MVA base. The transformer impedance is 0.354 + j6.92% on a 10MVA base,
and it is connected delta on the 138kV side, and grounded-wye on the 25kV side. The retained
25kV system has 109 busses. One-half of the combined line charging on the 138kV side of the
substation transformer is 13.5MVAr.
Page 8-11
June 2006
Plots of the positive/negative and zero sequence impedances at the television 25kV bus are
shown in Figure 8.5. It is clear that the 25kV system has parallel resonance for zero sequence
near the 3rd harmonic. With the delta, grounded-wye service transformer, the saturable
transformer does not “see” this 3rd harmonic resonance, but rather the relatively low impedance
of the transformer to ground.
Zero Seq Pos/Neg Seq

3.5
3
Log10 (Impedance - %)
2.5
1.5
0.5
0
1 3 5 7 9 11 13 15 17 19 21 23 25
Harmonic
Figure 8.5. Thevenin Equivalent

q Impedance
p vs. Frequency at 25kV
Television Station Bus
Summarizing, g, while the control instabilityy problem
p p
is not explained, , the simulations do confirm
that 3rd harmonic resonance coincides with the appearance
pp of the pproblem, and that by using a
delta, grounded-wye transformer, the situation can be returned to normal.
Page 8-12
June 2006
Case 4. 12.5kV Ski Area with 5150HP of Distributed Lift Motor ASDs
Overview. This case deals with the proposed expansion of a ski area. The 12.5kV underground
system will eventually have eight ski lifts powered byy DC motor drives,, totaling
g 5150HP. Total
load ((linear pplus nonlinear)) will be about 9MW. The DC motors will be driven by six-pulse line-
commutated ASDs so that the lifts will have soft-start, soft-stop operation. Measurements of the
proposed system are, or course, not possible. Thus, the harmonics situation must be analyzed in
advance using simulations.
Simulations. (PCFLOH Files *_SKIA.csv, *_SKIB.csv, *_SKIC.csv). A diagram of the ski

area is shown in Figure 8.6. In addition to the ASD loads, the ski area has 6 MVA of linear load.
The ASDs are modeled using the 1/k rule for harmonics through the 25th, with no phase angle
diversity. The dpfs of the ASDs and linear load are assumed to be 0.85. Cable capacitance is
taken from Table 8.5.
Table 8.5. Capacitance and Charging of 12.47kV Cables

Cable Capacitance kVAr
1/0 0.163 9.56
4/0 0.222 13.02
350 kcmil 0.267 15.65
500 kcmil 0.304 17.82
1000 kcmil 0.401 23.5
Capacitance: μF per km per phase

kVAr: (three-phase) per km
The point of common couplingg (PCC)

(PCC)
P is Bus #20, Substation 138 kV. Isc andd Iload d at the PCC
are 34.4 pu and 1.080 pu, respectively, on a 10MVA base. Twelve-month averagee Iload d is
estimated to be 0.75 • 1.080 = 0.810 pu, so Isc/Iloadd at the PCC is 42.5, and the corresponding
IEEE 519 limit for TDD of current is 8.0%.
The three cases studied are
Case SKIA. No Corrections.
Case SKIB. 30° phase shifting transformers added at Apollo and BigBoss ASDs.
Case SKIC. Case SKIB, plus 1800kVAr of filters.
Bracketed values in Figure 8.6 give solved THDV s for [Case SKIA, Case SKIB, Case SKIC],
except at the substation transformer, where THD I is given directly under Z.
The results for Case SKIA are shown in Figures 8.7 – 8.11. The highest voltage distortion is an
unacceptable 14.1% at Bus #12, Apollo.
Page 8-13
June 2006
For Case SKIB,, wye-delta
y transformers are added at approximately
pp y one-half of the ASD HP, so
that a net twelve-pulse operation for the entire ski area is approximated. Results are shown in
Figures 8.12 - 8.14. The highest voltage distortion reduces to 9.6% at Bus #12, Apollo.
Case SKIC builds upon Case SKIB by adding the following passive filters:
x 300kVAr of 5th at Bus#6, Base.

x 300kVAr of 5th at Bus#10, Taylor.
x 300kVAr of 7th at Bus#10, Taylor.
x 300kVAr of 11th at Bus#12, Apollo.
x 300kVAr of 11th at Bus#15, BigBoss.
x 300kVAr of 13th at Bus#13, Jupiter.
Filter X/R equals 50. The 5th and 7th harmonics have only one-half of the dedicated kVArs
because the two wye-delta transformers have already reduced 5th and 7th harmonic voltages.
Some 5th and 7th filtering is still needed in case one or both of the wye-delta transformers are out
of service (simulations for this contingency were made but are not presented here).
Results for Case SKIC are shown in Figures

g 8.15 – 8.18. The highest feeder voltage distortion
level falls to 2.9%, occurring at Bus #11, Longs.
A side benefit of the filters is that theyy correct the ski area power factor from 0.83 to 0.91, thus
providing both a harmonics and power factor solution.
Page 8-14
June 2006
#20, Substation 138kV,

V = 1.02 pu
Isc = 34.4 pu, [3.6, 2.4, 0.8] Disp. Power Factors
10 MVA base, Case SKIA = 0.83
X/R =5.0 Case SKIB = 0.83
Case SKIC = 0.91
#14, WipeOut (400 HP) #15, BigBoss (1250 HP)

[13.7, 9.3, 3.0] [14.0, 9.5, 2.9] 10/12.5/14 MVA Trans., Z = 6.54%, 10
MVA Base, X/R = 10, Load side tap =
1.025, Delta (138)-GWye (12.47),
THDI = [12.6, 5.7, 1.2]
#13, Jupiter
(1200 HP + 0.5 MVA) 1100m. 1/0AL 200m. #1, Substation 12.47kV
[13.8, 9.4, 2.8] 1000MCM,
1300m. 1/0AL [12.2, 8.3, 2.6]
2 ckts
600m. 1/0AL 800m. 350CU #2, Near Sub S.
#3, Near Sub N.
#8, Wilderness (250 HP)
#16, Shop
#12, Apollo #7, Star
(1200 HP) [14.1, 9.6, 2.8] 900m. 1/0AL
700m. 350CU 500m.
1000MCM,
300m. 350CU, 500m. 350CU #4, PBS 2 ckts
2 ckts (0.5 MVA)
#5, PBN
#10, Taylor
1000m. 350CU (0.5 MVA)
(400 HP + 1.0 MVA)
[13.4, 9.1, 2.8] #9, Dorsey
(250 HP) 500m. 1000MCM,
[13.3, 9.1, 2.8] 2 ckts
1200m. 350CU
1300m. 1/0AL Cable R+ X+
#6, Base Impedances (ohms/km) (ohms/km)
SKIA SKIB SKIC (3.0 MVA) 1000MCM 0.0929 0.1138
#11, Longs [12.8, 8.7, 2.7] 350 CU 0.1582 0.1228
(200 HP + 0.5 MVA)
1/0 AL 0.6460 0.2930
[13.5, 9.2, 2.9]
Figure 8.6: Case 4, Ski Area
Page 8-15
June 2006
Figure 8.7. Case SKIA Interface Screen

Note - Solved THDV s are shown. The highest THDV , 14.1% at Bus #12, Apollo, is identified by the asterisk.
Page 8-16
June 2006
Deep notching
o ( THDV = 14.1%)
Figure 8.8. Case SKIA, Voltage Waveform and Spectrum at Bus #12, Apollo
o ( THD I =
Figure 8.9. Case SKIA, Current Injection Waveform and Spectrum at Bus #12, Apollo
29.0%)
Page 8-17
June 2006
Figure 8.10. Case SKIA, Voltage Waveform and Spectrum at Bus #1, Substation 12.5 kV
( THDV = 12.2%)
Figure 8.11. Case SKIA, Substation Transformer Current Waveform and Spectrum on 138 kV
Side ( THD I = 12.6%)
Page 8-18
June 2006
Figure 8.12. Case SKIB, THDV
Effect of net 12-pulse

operation. More
notches, but smaller.
Figure 8.13. Case SKIB, Voltage Waveform and Spectrum at Bus #6, Base ( THDV = 8.7%)
Page 8-19
June 2006
Figure 8.14. Case SKIC, THDV
Figure 8.15. Case SKIC, Voltage Waveform and Spectrum at Bus #6, Base ( THDV = 2.7%)
Page 8-20
June 2006
Figure 8.15. Case SKIC, Substation Transformer

f Current Waveform and Spectrum on 138kV
Side ( TH
THD
HD I = 1.2%)
Page 8-21
Grady Chapter 9. Homework Problems Page 9-1
June 2006
Problem 1 – Partial Harmonic Current Cancellation Due to Variations in DPF.
Two identical six-pulse phase adjustable-speed motor drives operate side by side. Each draws
500kVA from a 480V bus. One drive has DPF = 0.85, and the other has DPF = 0.75. Assume that
x the two drive current waveforms have the same standard six-pulse waveform, except that one
lags the other in time according to their DPFs.
x the 5th harmonic current for each drive is 20% of fundamental
What is the net 5th harmonic current drawn by the pair, in rms amperes?
Problem 2 – Estimating the Filter kVArs Needed to Correct a Seriously Distorted Distribution
System.
Highly-distorted distribution systems, such as ski areas with large six-pulse adjustable-speed motor
drives, require significant filtering. Conventional shunt capacitors must be avoided – they must be
used only as filters. A good “starting point” for developing a filter plan is to:
a. power factor correct the distribution system DPF to 0.95 with harmonic filters,
b. use one-half as many kVArs for 7th harmonic filters as you do for 5th,
c. assume that contributions of harmonics above the 7th will not be a serious problem.
If a distribution system has 10MW, 5MVAr of conventional load, and 3MW of six-pulse drives
(assume drive DPF = 0.85), determine a “starting point” estimate for the net kVArs needed for 5th, and
also for 7th, harmonic filters.
Problem 3 í Harmonic Filter Design.
A 12.5kV distribution feeder is experiencing a problem with high 5th harmonic voltages caused by
large six-pulse adjustable-speed motor drives. Your task is to design a “generic” 5th harmonic,
grounded-wye, series-tuned filter. Each phase of the filter will employ a 200kVAr capacitor and a
series inductor. Ignore inductor resistance. (Note - the number of these generic filters needed to solve
the harmonic problem is not addressed here)
a. Specify the inductance (mH) of each inductor.

b. Determine the 60Hz rms current in each inductor.
Now, assuming that the filter will have a 5th harmonic current with rms magnitude equal to the
fundamental,
c. Specify the rms ampere rating of each inductor.

d. Determine the rms 5th harmonic voltage that will appear across each capacitor.
Page 9-1
Grady Chapter 9. Homework Problems Page 9-2
June 2006
Problem 4 - Filter Design for Ski Area of Case 4 in Chapter 8, “Case Studies.”
Step 1. Data files bdat_skia.csv and ldat_skia.csv were prepared on a 10MVA base and using the
information given in Case 4. Examine the .csv files and, by comparing their data to the Case 4
description, do the following:
x Verify that the R, X, and line charging VAr information in ldat_skia.csv for the segment
between Bus 6 (Base) and Bus 7 (Star) is correct.
x Verify that the nonlinear load information in bdat_skia.csv for Bus 14 (WipeOut) is correct.
x Verify that the 138kV transmission system equivalent information given in bdat_skia.csv
agrees with the information given in the Case 4 description (i.e., the subtransient values in the
file are derived from the Isc = 34.4pu, 10MVA base, X/R = 5.0 values shown in the box).
Step 2. Run PCFLOH using the given bdat_skia.csv and ldat_skia.csv files and confirm that you
obtain the same results given in Figure 8.6 and in the spectral contents given in revised Figures 8.7-
8.11. To run PCFLOH.exe, bdat_skia.csv, and ldat_skia.csv in a directory, click on PCFLOH.exe, and
enter _skia.csv in the input field. Follow the instructions by clicking the buttons. View waveforms by
clicking on bus numbers or branches.
Step 3. Without using phase shifting transformers, design and test a filter strategy that meets the
following criteria:
x Displacement power factor at the substation transformer 138kV bus is corrected so that it falls
between 0.95 (lagging) and 1.0.
x Filters are either 300kVAr or 600kVAr each.
x Tune 0.3pu Hz (i.e., 0.3 Ɣ 60 = 18Hz) below the harmonic target.
x Assume X(at 60Hz)/R ratio of filter inductors = 50.
x No more than 600kVAr of filters are added at any single bus.
x Max THDv in the 12.5kV system < 4.5% (ignore the contributions of harmonics above the 13th)
x Max fundamental V1 in the 12.5kV system, excluding filter capacitors, is less than 105% (note
– see Step 4 if you cannot achieve this)
Use impedance scans to confirm that the filters produce impedance dips at the desired frequencies.
Note – a filter example is given in Chapter 6, “Conducting an Investigation.”
Step 4. If you are unable to satisfy the V1 limit, then consider adding a phase shifting transformer to a
large drive. This can be done by specifying a “nonlinear device phase shift” of 30 degrees in the
corresponding row of the BDAT file. It may be necessary to remove some of the filters.
Step 5. Your final BDAT and LDAT files, containing your solution, should be named
bdat_yourname.csv, ldat_yourname.csv, and emailed to Dr. Grady for checking.
Step 6. Prepare a final report (no more than 3 pages) for your client.
Page 9-2
Grady Chapter 10. Acknowledgements Page 10-1
June 2006
Manyengineers
Many engineershave
havecontributed
contributedtotothis
thisdocument,
document,either
eitherthrough
throughinspiration,
inspiration,examples,
examples,oror
advice.InInparticular,
advice. particular,I would
and in like
no particular order, IMansoor,
to thank Arshad would likeRayto thank Arshad
Stratford, Mansoor,
Michael Ray
Doyle, John
Stratford, Michael Doyle, John Soward, Scott Jackson, Diane Ammons, Ewald
Soward, Scott Jackson, Diane Ammons, Ewald Fuchs, David Hartmann, Dennis Hansen, Russell Fuchs, David
Hartmann,
Ehrlich, BlaineDennis Hansen,
Leuschner, Russell
John Ehrlich,
Soward, andNarendorf,
Martin Blaine Leuschner.
and Rao Thallam.
Page 10-1
Grady Appendix A1. Fourier Spectra Data Page A1-1
June 2006
Spectral Data for Figure 2.1, Spectral Data for Figure 3.3 (Top),
120V PC Current Idealized Six-Pulse Current-Source ASD Current
Ang - Ang - Ang - Ang -
Harmonic Mag - % Sine Deg. Harmonic Mag - % Sine Deg. Harmonic Mag - % Sine Deg. Harmonic Mag - % Sine Deg.
1 100.00 5 2 1 100.00 0 2
3 88.90 -177 4 3 4
5 73.60 5 6 5 20.00 180 6
7 54.40 -173 8 7 14.29 180 8
9 35.60 9 10 9 10
11 19.00 -167 12 11 9.09 0 12
13 7.40 20 14 13 7.69 0 14
15 1.00 -88 16 15 16
17 2.80 180 18 17 5.88 180 18
19 3.20 13 20 19 5.26 180 20
21 2.40 -157 22 21 22
23 1.90 43 24 23 4.35 0 24
25 1.80 -119 26 25 4.00 0 26
27 1.90 67 28 27 28
29 2.00 -110 30 29 3.45 180 30
31 1.60 75 32 31 3.23 180 32
33 1.00 -85 34 33 34
35 0.80 160 36 35 2.86 0 36
37 1.20 23 38 37 2.70 0 38
39 1.50 -140 40 39 40
41 1.30 54 42 41 2.44 180 42
43 0.70 -111 44 43 2.33 180 44
45 0.50 142 46 45 46
47 0.70 15 48 47 2.13 0 48
49 0.80 -157 50 49 2.04 0 50
Page A1-1
June 2006
Spectral Data for Figure 3.3 (Bottom Left), Spectral Data for Figure 3.3 (Left),
Idealized Six-Pulse Current-Source ASD Current with 30° Shift Idealized Twelve-Pulse Current-Source ASD Current
1 100.00 0 2 1 100.00 0 2
3 4 3 4
5 20.00 0 6 5 6
7 14.29 0 8 7 8
9 10 9 10
11 9.09 0 12 11 9.09 0 12
13 7.69 0 14 13 7.69 0 14
15 16 15 16
17 5.88 0 18 17 18
19 5.26 0 20 19 20
21 22 21 22
23 4.35 0 24 23 4.35 0 24
25 4.00 0 26 25 4.00 0 26
27 28 27 28
29 3.45 0 30 29 30
31 3.23 0 32 31 32
33 34 33 34
35 2.86 0 36 35 2.86 0 36
37 2.70 0 38 37 2.70 0 38
39 40 39 40
41 2.44 0 42 41 42
43 2.33 0 44 43 44
45 46 45 46
47 2.13 0 48 47 2.13 0 48
49 2.04 0 50 49 2.04 0 50
Page A1-2
June 2006
Spectral Data for Figure 4.1, Spectral Data for Figure 4.2,
25kV/240V Transformer Magnetizing Current 120V Refrigerator Current
1 100.00 -66 2 1 100.00 -31 2 3.50 -48
3 63.50 -69 4 3 5.10 101 4 0.40 98
5 35.90 -70 6 5 0.50 124 6 0.30 152
7 18.30 -69 8 7 1.00 10 8 0.20 55
9 10.10 -73 10 9 0.40 -57 10
11 5.40 -78 12 11 0.20 -103 12
13 2.10 -80 14 13 0.20 -180 14
15 0.90 -101 16 15 0.10 129 16
17 0.40 -96 18 17 18
19 0.10 -90 20 19 20
21 22 21 22
23 0.20 -91 24 23 24
25 0.20 -174 26 25 26
27 0.10 126 28 27 28
29 30 29 30
31 0.20 21 32 31 32
33 0.20 99 34 33 34
35 36 35 36
37 0.10 -44 38 37 38
39 40 39 40
41 42 41 42
43 0.30 -90 44 43 44
45 46 45 46
47 0.40 64 48 47 48
49 0.20 -165 50 49 50
Page A1-3
June 2006
240V Residential Air Conditioner Current 277V Fluorescent Lamp Current (with Magnetic Ballast)
1 100.00 -23 2 1 100.00 -2 2 2.00 -117
3 8.00 110 4 3 15.80 -150 4 0.40 168
5 6.80 58 6 5 8.60 -128 6 0.30 47
7 0.50 165 8 7 2.90 -73 8 0.10 8
9 0.60 82 10 9 2.00 -32 10
11 0.30 157 12 11 1.40 6 12
13 14 13 0.80 29 14 0.20 -49
15 16 15 0.40 32 16
17 0.20 87 18 17 0.20 -17 18
19 20 19 0.50 -56 20
21 22 21 0.40 -49 22 0.20 -133
23 24 23 0.20 -65 24
25 26 25 26
27 28 27 0.10 162 28
29 30 29 30
31 32 31 32
33 34 33 34 0.10 -166
35 36 35 0.10 -42 36
37 38 37 38
39 40 39 40 0.10 149
41 42 41 42
43 44 43 44
45 46 45 46
47 48 47 0.10 -137 48 0.20 25
49 50 49 50
Page A1-4
June 2006
Spectral Data for Figure 4.6 (Top Left), Spectral Data for Figure 4.6 (Top Right),
Six-Pulse Voltage-Source ASD Current (High Power) Six-Pulse Voltage-Source ASD Current (Low Power)
1 100.00 0 2 1 100 0 2
3 4 3 4
5 29.81 180 6 5 58.89 180 6
7 3.22 180 8 7 30.78 0 8
9 10 9 10
11 8.95 0 12 11 5.3 0 12
13 3.12 0 14 13 8.64 180 14
15 16 15 16
17 5.05 180 18 17 0.34 0 18
19 2.52 180 20 19 3.04 0 20
21 22 21 22
23 3.48 0 24 23 1.43 180 24
25 2.07 0 26 25 0.85 180 26
27 28 27 28
29 2.64 180 30 29 1.48 0 30
31 1.75 180 32 31 0.14 180 32
33 34 33 34
35 2.13 0 36 35 1.2 180 36
37 1.51 0 38 37 0.55 0 38
39 40 39 40
41 1.78 180 42 41 0.84 0 42
43 1.33 180 44 43 0.67 180 44
45 46 45 46
47 1.53 0 48 47 0.49 180 48
49 1.18 0 50 49 0.61 0 50
Page A1-5
June 2006
Spectral Data for Figure 4.6 (Bottom Left), Spectral Data for Figure 4.6 (Bottom Right),
Six-Pulse Voltage-Source ASD Current (High Power) Six-Pulse Voltage-Source ASD Current (Low Power)
with 30° Shift with 30° Shift
1 100.00 0 2 1 100.00 0 2
3 4 3 4
5 29.81 0 6 5 58.89 0 6
7 3.22 0 8 7 30.78 180 8
9 10 9 10
11 8.95 0 12 11 5.30 0 12
13 3.12 0 14 13 8.64 180 14
15 16 15 16
17 5.05 0 18 17 0.34 180 18
19 2.52 0 20 19 3.04 180 20
21 22 21 22
23 3.48 0 24 23 1.43 180 24
25 2.07 0 26 25 0.85 180 26
27 28 27 28
29 2.65 0 30 29 1.48 180 30
31 1.75 0 32 31 0.14 0 32
33 34 33 34
35 2.13 0 36 35 1.20 180 36
37 1.51 0 38 37 0.55 0 38
39 40 39 40
41 1.78 0 42 41 0.84 180 42
43 1.33 0 44 43 0.67 0 44
45 46 45 46
47 1.53 0 48 47 0.49 180 48
49 1.18 0 50 49 0.61 0 50
Page A1-6
June 2006
120V Microwave Oven Current 120V Vacuum Cleaner Current
1 100.00 -3 2 11.60 -102 1 100.00 -18 2 0.40 97
3 27.80 -149 4 1.10 97 3 25.70 132 4 0.70 -118
5 9.50 -61 6 1.50 -109 5 2.70 -70 6 0.20 79
7 3.30 -46 8 7 1.80 -126 8 0.30 -74
9 1.90 45 10 0.30 -78 9 0.40 112 10
11 1.10 139 12 0.20 -21 11 0.60 -33 12
13 0.90 -144 14 0.10 68 13 14
15 0.60 -60 16 0.10 156 15 0.20 123 16
17 0.50 17 18 17 18
19 0.40 88 20 19 20
21 0.30 162 22 21 22
23 0.30 -121 24 23 24
25 0.30 -45 26 25 26 0.10 63
27 0.20 25 28 27 28
29 0.20 103 30 29 30
31 0.20 -174 32 31 32
33 0.20 -111 34 33 0.20 147 34
35 0.20 -47 36 35 0.20 -76 36
37 0.10 39 38 37 38
39 0.20 114 40 39 40
41 0.10 -7 42 41 42
43 0.10 -126 44 43 44
45 0.10 -43 46 45 46
47 48 47 48
49 50 49 50
Page A1-7
June 2006
277V Fluorescent Lamp Current (with Electronic Ballast) Voltage at 120V Service Panel for PC Workstations
1 100.00 4 2 0.20 32 1 100.00 0 2
3 7.30 -166 4 0.40 111 3 2.20 53 4
5 8.00 -1 6 0.30 84 5 3.90 -150 6
7 3.00 -3 8 0.20 67 7 1.40 -28 8
9 0.20 -85 10 0.10 86 9 0.60 54 10
11 1.40 -165 12 0.10 60 11 0.90 -176 12
13 1.10 176 14 0.30 60 13 0.50 -60 14
15 0.90 -168 16 0.20 -65 15 0.50 25 16
17 0.40 -164 18 0.10 19 17 0.50 139 18
19 0.30 -59 20 19 0.30 -126 20
21 0.70 -32 22 0.20 61 21 0.30 -12 22
23 0.30 -39 24 0.30 162 23 0.30 112 24
25 0.40 -54 26 0.30 -87 25 0.30 -155 26
27 0.20 -39 28 0.40 95 27 0.40 -47 28
29 0.20 -73 30 29 0.40 38 30
31 0.30 -136 32 0.10 -81 31 0.40 134 32
33 0.60 -128 34 33 0.30 -118 34
35 0.40 -126 36 0.10 0 35 0.40 -44 36
37 0.30 -149 38 0.10 148 37 0.20 78 38
39 0.20 -112 40 0.20 102 39 0.40 161 40
41 0.50 -130 42 41 0.40 -102 42
43 0.40 -140 44 0.20 93 43 0.40 -39 44
45 0.60 156 46 0.20 51 45 0.20 35 46
47 0.30 177 48 0.20 -130 47 0.20 163 48
49 0.30 177 50 49 0.20 -127 50
Page A1-8
June 2006
Spectral Data for Figure 5.10,

480V Service Entrance Current for College of Business
Administration, U. T. Austin
Ang - Ang -
Harmonic Mag - % Sine Deg. Harmonic Mag - % Sine Deg.
1 100.00 -27 2 0.10 0
3 3.50 -91 4 0.10 -42
5 4.50 -146 6 0.00 0
7 3.70 -48 8 0.10 61
9 0.40 41 10 0.10 145
11 1.00 10 12 0.10 -30
13 0.70 138 14 0.10 -134
15 0.00 0 16 0.20 38
17 0.40 -104 18 0.00 0
19 0.30 -80 20 0.00 0
21 0.00 0 22 0.00 0
23 0.20 -153 24 0.00 0
25 0.00 0 26 0.10 54
27 0.00 0 28 0.20 173
29 0.40 -165 30 0.10 -67
31 0.20 141 32 0.00 0
33 0.00 0 34 0.00 0
35 0.00 0 36 0.10 -162
37 0.00 0 38 0.00 0
39 0.10 143 40 0.10 6
41 0.20 93 42 0.20 58
43 0.00 0 44 0.20 -99
45 0.30 126 46 0.10 176
47 0.00 0 48 0.00 0
49 0.20 176 50 0.10 169
Page A1-9
Grady Appendix A2. PCFLO User Manual Page A2-1
June 2006
PCFLO User Manual

(Including PCFLO and PCFLOH)
June 2006
Prof. Mack Grady
Department of Electrical & Computer Engineering
University of Texas at Austin
Austin, TX 78712
Phone: 512 - 471 - 5231
FAX: 512 - 475 - 8198
email: grady@mail.utexas.edu
1. Introduction
PCFLO performs loadflow, short circuit, and harmonics analysis. It reads column-formatted,
comma-separated (CSV), or tab-separated (TSV) data. CSV/TSV data make it easy to use
Microsoft Excel for preparing your input data and examining your output results. Excel’s
sorting capabilities are especially helpful in reviewing, for example, total harmonic distortion at
busses, or in finding transmission lines with overloads or busses with high or low voltages.
There are actually two versions of PCFLO.
x Regular PCFLO performs loadflow, short circuit, and harmonics analysis for systems as
large as 500 busses. It has a the interface screen show below in Figure 1. It reads input files
BDAT, LDAT, ADAT, OPTIONS, (and for harmonics, file SPECTRA) that have the path
and extension entered on the screen. For short circuit studies, you can enter the faulted bus
Page A2-1
June 2006
and fault impedance on the screen. The information in file OPTIONS instructs PCFLO to
perform loadflow (including motor starting), short circuit, or harmonics calculations.
Figure 1. PCFLO Interface Screen

(Note – specify input/output file path and extension here)
x PCFLOH is a 500 bus harmonics-only version of PCFLO, with the interface screen shown in
Figure 2. As with PCFLO, it reads the BDAT, LDAT, OPTIONS, SPECTRA input files
with the path and extension specified.
Excel-Visual Basic programs PCFLO_VSOLN_PLOT, PCFLO_ISOLN_PLOT, and

PCFLO_ZBUS_PLOT have been added to produce Excel plots. When executing, click the
“Enable Macros” button.
2. Author and Contact
Professor Mack Grady, Department of Electrical & Computer Engineering, University of Texas
at Austin, Austin, TX 78712. Phone: 512-471-5231, FAX: 512-475-8198, email:
grady@mail.utexas.edu.
Page A2-2
June 2006
Figure 2. PCFLOH Interface Screen

(Note – specify input/output file path and extension, plus solution options and filters here)
Page A2-3
June 2006
3. Setup Instructions and Introduction
You will find that HOWDY.ZIP contains some column- formatted files, and some comma-
separated or tab-separated (CSV/TSV) data files. Four CSV header files are also enclosed that
you may find helpful when using Excel (files BDAT.HDR, LDAT.HDR, ADAT.HDR,
SPECTRA.HDR). These header files should be placed at the top of CSV/TSV input data to
describe the content of the spreadsheet columns.
For your information, PCFLO treats any row in a CSV/TSV input file that has a colon in column
1 as a comment line.
Setup Instructions:
1. Establish a new directory named C:\PCFLO,
2. Download HOWDY.ZIP into it,
3. Pkunzip HOWDY.ZIP into C:\PCFLO.
4. Running a Sample Harmonics Case with PCFLOH
To run a sample case with PCFLOH, follow these instructions after you have unzipped
HOWDY.ZIP into directory C:\PCFLO
x Double-click the PCFLOH icon. The interface screen shown in Figure 2 will appear.
x In Step 1 of the PCFLOH screen, click Check Input and then click Check Output. The
default names in the entry boxes provide an example case. Later, you can change these
names to match your own file names or paths.
x In Steps 2,3,4, consecutively, click Continue. Note - abbreviated instructions are given in
the Program Message box at the top right of the PCFLOH screen.
x In Step 5, you can view the linear load, nonlinear load, and capacitor specifications at
individual busses by clicking the corresponding names in the bus list on the right of the
screen. You can also add filters. When ready to continue, click Continue to Step 6.
x In Step 6, click Solve. Once the case solves, click any of the busses shown in the list on the
right to see its voltage waveform or spectrum. Click on I to repeat this procedure for branch
currents. Waveforms are shown in the background. The red curve is the fundamental
component. The max/min y-scale on the graph corresponds to the sum of fundamental and
harmonic magnitudes for the waveform.
5. Instructions for Performing Loadflow, Short Circuit, Harmonics, and Motor Starting
Calculations
The information in file OPTIONS instructs PCFLO to perform loadflow (including motor
starting), short circuit, or harmonics calculations.
Input data for PCFLO can be either column formatted, comma-separated, or tab-separated. Most
output files are formatted and comma-separated. Comma-separated and tab-separated data files
are easily viewed and edited with Microsoft Excel.
Page A2-4
June 2006
PCFLO always reads data directly from files ADAT, BDAT, LDAT, OPTIONS, and SPECTRA.
Their content and structures are explained in this document.
5a. Loadflow
Step 1: Copy file OPTIONS.LF to your file OPTIONS and make any necessary changes.
Step 2: Copy the loadflow bus and line/transformer into files BDAT and LDAT, respectively.
If area interchange data are desired, include them in file ADAT. Otherwise,
ADAT is unnecessary and must be deleted.
Step 3: Execute PCFLO
Step 4: The loadflow "printout" is found in file OUT1 and OUT4. File VSOLN contains the
solved bus voltages. File ISOLN contains branch currents. File OUT5 contains a
useful summary of both branch currents and bus voltages.
5b. Short Circuit (PCFLO)

(Note – You can use file extensions with PCFLO)
Step 1: Copy file OPTIONS.SC to your file OPTIONS and make any necessary changes.
Step 2: BDAT must contain the conventional loadflow data plus, at a minimum, the positive
sequence subtransient reactances of the swing bus and other generators and motors.
If negative and zero sequence data are provided for generators and motors, including
neutral grounding impedances, then PCFLO will be able to perform line-to-line and
line-to-ground fault calculations (as well as three-phase fault calculations).
Similarly, LDAT must contain conventional loadflow data. Also, if line-to-line and
line-to-ground fault calculations are to be made, then LDAT must include negative
and zero sequence data, such as transformer connections, phase shifts, and neutral
grounding impedances.
Step 3: Execute PCFLO. The "printout" is found in files OUT1 and OUT2. PCFLO builds
impedance matrices ZBUS0, ZBUS1, and ZBUS2 for zero/positive/negative
sequences, respectively. ZBUS1 is always built, but ZBUS0 and ZBUS2 are built
only if sufficient data are provided in BDAT and LDAT.
While in PCFLO, enter the bus number and complex fault impedance, in percent.
Enter as many combinations of bus numbers and fault impedances as you like, one at
a time.
Step 4: The results are found in file FREP. FREP lists all the information for a three-phase
fault, then a phase-b-to-c line-to-line fault (through Z F ), then a phase-a line-to-
ground fault (through Z F ).
Page A2-5
June 2006
5c. Full Harmonic Study (PCFLO or PCFLOH)

(Note – You can use file extensions with PCFLO or PCFLOH)
A full harmonic study injects harmonic load currents and calculates harmonic network bus
voltages, branch currents, and voltage distortion.
Step 1: Copy file OPTIONS.HS to your file OPTIONS and make any necessary changes.
Step 2: BDAT and LDAT must contain the same data needed for fault studies. In addition,
BDAT must contain the characteristics of any connected nonlinear loads.
Step 3: User-specified harmonic current spectra for nonlinear loads can be placed into file
SPECTRA if needed.
Step 4: Execute PCFLO (or PCFLOH). A loadflow is automatically invoked to help identify
data errors.
Step 5: The "printout" is found in files OUT1, OUT2, and OUT3. PCFLO builds files
VSOLN and ISOLN that contain the Fourier series of bus voltages,
line/transformer/shunt element currents, and nonlinear device currents (load
convention). The values are given in rms percent. PCFLO also builds file THDV
which tabulates network voltage distortions.
Step 6: Excel-Visual Basic programs PCFLO_VSOLN_PLOT and PCFLO_ISOLN_PLOT

have been added to produce Excel plots of bus voltages, nonlinear load currents, and
branch currents. When executing, click the “Enable Macros” button.
5d. Harmonic Impedance Scan (PCFLO or PCFLOH)

(Note – You can use file extensions with PCFLO or PCFLOH)
A harmonic impedance scan computes the Thevenin equivalent (i.e. driving point) harmonic
impedances of network busses.
Step 1: Copy file OPTIONS.HZ to your file OPTIONS, and make the necessary changes
according to Table 4.
Step 2: BDAT and LDAT must contain the same data needed for fault studies. In addition,
BDAT must contain the characteristics of any connected nonlinear loads.
Step 3: Execute PCFLO (of PCFLOH). A loadflow is automatically invoked for either
option to help identify data errors.
Step 4: The "printout" is found in files OUT1, OUT2, and OUT3. PCFLO builds the
diagonal elements of either file ZBUS0, ZBUS1, or ZBUS2, depending on which
Page A2-6
June 2006
phase sequence is requested. The values are given in percent. If desired, impedance
scans can be limited to nonlinear busses.
Step 5: Excel-Visual Basic program PCFLO_ZBUS_PLOT has been added to produce Excel
plots of impedance matrix element impedance magnitudes and angles. When
executing, click the “Enable Macros” button.
5e. Motor Starting
Step 1: Copy file OPTIONS.LF to your file OPTIONS, and make the necessary changes
according to Table 1.
Step 2: Copy the corresponding loadflow bus and line/transformer into files BDAT and
LDAT, respectively. If area interchange data are desired, include them in file ADAT.
Otherwise, ADAT should be deleted.
Step 3: Modify file LDAT so that the motor bus has a fixed series impedance-to-ground
branch. The FROM BUS of the branch is the motor bus number, and the TO BUS is
zero (i.e., ground). Since the starting current for a motor is approximately five-times
the full-load run current, and at a very low power factor, the impedance should be
approximately:
ZM R M jX M ,
where
pf XM
X M | , R M | ,
5 x P M 5
and
P M , pf
are full-load motor power and power factor, respectively. The impedance should be
entered in percent.
Step 4: Remove the motor P and Q load from file BDAT.
Step 5: Execute PCFLO. The results are found in files OUT4 and VSOLN. If the voltage at
the motor bus falls below 80%, it may not start.
6. Summary of PCFLO and PCFLOH Data Files
ADAT: Loadflow area input data. (PCFLO also creates temporary file ADAT.TMP)
ASOLN: Solved area data for loadflows.
Page A2-7
June 2006
BDAT: Bus input data.
BDAT.TMP: Unformatted bus data file built by PCFLO and read by FAULTS.
BORDER: File built by PCFLO that lists the busses in optimal order.
EXLOG: Echo print of screen messages.
FREP: Output for short circuit studies (produced by program FAULTS).
ISOLN, VSOLN: Solved branch currents and bus voltages for loadflow and harmonics
studies. (Sine series format for Fourier series)
LDAT: Line and transformer input data.
LDAT.TMP: Unformatted line and transformer data file built by PCFLO and read by
FAULTS.
OPTIONS: User input solution options.
OUT1, OUT2, OUT3: Echo print of input data for loadflow, short circuit, and harmonics,
along with pertinent messages and errors.
OUT4: Loadflow output data.
OUT5: Loadflow summary output used for wheeling analysis.
SPECTRA: User input harmonic current injection spectra. (PCFLO also creates temporary
file SPECTRA.TMP)
THDV: Solved total harmonic voltage distortions for harmonics studies.
ZBUS0, ZBUS1, ZBUS2: Solved zero/positive/negative impedance matrix elements for

short circuit (rectangular form) or harmonic studies (polar
form), in percent.
7. Prepared Input Data Files
(Note - always delete the old ADAT file before performing a loadflow study)
18 Bus Distribution System for Harmonics Studies. Extension WG.
83 Bus Transmission System for Loadflow Studies. Extension COA.
5 Bus Distribution System for Harmonics Studies. Extension EL.
Page A2-8
June 2006
5 Bus Stevenson* Loadflow Example, pp. 200-205. Extension S5.
6 Bus Grainger-Stevenson** Short Circuit Example (Prob. 3.12, p. 139, and continued with
Prob. 11.17, p. 469). Extension S6.
9 Bus Grainger-Stevenson** Short Circuit Example (Prob. 3.13, pp. 139-140, and continued
with Prob. 11.18, p. 469). Extension S7.
36 Bus Industrial Short Circuit Example. Extension DP1.
232 Bus Distribution System for Harmonics Studies. Extension CHI.
* William D. Stevenson, Jr., Elements of Power System Analysis, Fourth Edition, McGraw-
Hill, New York, 1982.
** John J. Grainger, William D. Stevenson, Jr., Power System Analysis, McGraw-Hill, New
York, 1994.
8. Structure of Harmonic-Related Output Files
(Note - for loadflow studies, the formats of ISOLN and VSOLN are different from below but are
self-explanatory when viewing the files. For short circuit studies, the ZBUS files are similar to
below, but written in rectangular form)
Commas separate the header lines and fields shown below to facilitate their use with Microsoft
Excel.
ISOLN
Data Field
(starting from
the left) Description
1 Harmonic number
2 From bus number
3 To bus number
4 Circuit number
5 Current magnitude - percent
6 Current phase angle (sine reference) - degrees
7 From bus name (at the first opportunity only)
8 To bus name (at the first opportunity only)
9 Loading level - % of line rating (for
fundamental frequency only)
Page A2-9
June 2006
VSOLN
Data Field
(starting from
1 Harmonic number
2 Bus number
3 Voltage magnitude - percent
4 Voltage phase angle (sine reference) - degrees
5 Nonlinear device load current magnitude -
percent
6 Nonlinear device load current phase angle
(sine reference) - degrees
7 Bus name (at the first opportunity only)
ZBUS0, ZBUS1, ZBUS2
Data Field
(starting from
1 Harmonic number
2 From bus number
3 To bus number
4 Impedance magnitude - percent
5 Impedance phase angle - degrees
6 From bus name (at the first opportunity only)
7 To bus name (at the first opportunity only)
THDV
THDV contains a list of bus numbers with their corresponding names and voltage
distortions.
9. Error Messages
*** error code 1 ***

either one of the bus numbers is out of range
or the from-to numbers are identical

number of lines has exceeded program limit = 1500

invalid transformer tap or tap range. acceptable
Page A2-10
June 2006
values are 0.50 to 1.50

bus number out of range - highest allowed number = 99999

bus type out of range

more than one bus specified as swing bus

no swing bus specified

too many busses - maximum number of busses = 500

this bus number has already been specified

one of these bus numbers was not included in the bus data

this line impedance is too small

negative resistances not allowed

there must be at least two busses to run the loadflow

the loadflow did not converge to the desired worst bus p,q mismatch
tolerance = .0005

the jacobian dimension (maxjac) must be at least twice as large as the number
of busses (nb). nb = 5, maxjac = 1000.

network not contiguous. bus 1, blank, has no path to bus 3, maple .

a transformer tap should not be specified for a shunt element
Page A2-11
June 2006
voltage should not be specified at a pq bus

this specified voltage is outside the normal range of 50% to 150%

phase shift should not be specified for a shunt element

the maximum number of jacobian storage vector elements is maxspr = 500000.
this length was exceeded for the storage command at row = 10,
column = 5.

this area is outside the acceptable range of 1 to 99

the area specified for bus 10 is not included
in the area interchange data

the total area power export should be zero

a control bus for this area has been specified already

this bus is not in the area that it will control

maxyb = 3500, nb = 5, nlines = 6. maxyb not big enough.

program error in ybuild when processing line 10 - 5
look in the section which examines knpt2

there are no lines or transformers in file ldat

q limits or remote voltage control should not be
specified at a pq bus

min q must not exceed max q
Page A2-12
June 2006

transformer information should not be specified for a non-transformer

transformer tap outside the specified limits

remote voltage control bus number out of range.
highest allowed bus number = 99999

remote voltage controlled bus number 10 does not
have a corresponding bus data entry

the swing bus must control its own voltage

remote controlled bus 10 should be a pq bus

conflicting voltages specified for bus 10

voltage specified, but no control bus given

the control side designator must be 1 or 2

program error in updating y matrix after tap changes

transformer control side given, but no control bus specified

transformer or shunt element connection type should not
be specified

transformer or shunt element grounding impedance should not
be specified

transformer grounding impedance should not be specified for
Page A2-13
June 2006
this type of transformer connection

the swing bus should have a subtransient impedance

grounding impedance should not be specified for this type
of connection

illegal transformer connection type

illegal connection type

the branch between busses 10 and 5 results in a loop
with net nonzero phase shift

phase shift too large

positive sequence short circuit impedance is zero, but short circuit
data for other sequences are nonzero

negative line charging not permitted

p generation or p load not permitted

nonlinear load,generation cannot be larger than total load,generation

nonlinear load displacement factor must be no worse than 0.1,
and no better than 1.0

nonlinear load type must be between 1, and 33, inclusively

illegal solution option in file = options
Page A2-14
June 2006
program options must be 1 for loadflow, 2 for short circuit, 3 for full harmonic study,
or 4 for harmonic impedance scan

the highest harmonic must be between 1 and 49, inclusively

enter 0 for zero sequence, 1 for positive sequence, or 2 for negative sequence

cannot perform zero sequence harmonics study or impedance study because
some zero sequence data are missing

cannot perform negative sequence harmonics study or impedance study because
some negative sequence data are missing

the nonlinear load must be either a load or a generator, not both

a full harmonics solution is requested, but there are no harmonics sources

negative subtransient reactances or grounding reactances not permitted

no user-specified spectral data for bus 10

program internal or user-specified fundamental spectral magnitudes must be 100.0

repeated user-specified nonlinear spectral data

must specify sin or cos

for any nonlinear load, the series types must all be sin or cos

if provided, skin effect resistance-doubling harmonic must be 2.0 or larger

invalid harmonics load model type specified for linear pq loads
Page A2-15
June 2006

nonlinear device at bus = 0, blank has zero sequence current but no zero sequence path to
ground

min and max scan harmonics must be between 1 and 49.
the min must not exceed the max.

scan steps per harmonic must be between 1 and 100

data format error while reading the options file

data format error while reading the vsoln or isoln file

data format error while reading the adat file
check line 1 (excluding any valid header lines),
make sure that area names having commas are enclosed in quotations.

data format error while reading the spectra file
check line 1 (excluding any valid header lines)

data format error while reading the bdat file
check line 6 (excluding any valid header lines),
make sure that bus names having commas are enclosed in quotations.

data format error while reading the ldat file
check line 7 (excluding any valid header lines)

data format error while reading the optord file

either no scan busses have been specified, or a specified scan bus
is not in the bus data file

phase regulating transformers can be used only in loadflow cases
Page A2-16
June 2006
phase regulating range outside -180 to +180 degrees

phase shift specified is outside the phase regulating range given

the only nonlinear bus continuation data permitted
are bus number and nonlinear load parameters

more than 10 nonlinear loads at this bus

invalid built-in nonlinear load type 10

global resistance doubling frequency 0 must be either zero (to ignore it)
or greater than 1
(
Page A2-17
June 2006
Important Changes Since PCFLO Version 3.7: Linear and Nonlinear Components of Load and Generation at a Bus are Now
Specified Entirely Separately, as Highlighted in Bold Type. Also, Nonlinear DPF is now in pu instead of %)
BUS DATA
(File = BDAT, one record per bus. May use CSV/TSV format if preferred)
Variable Columns Format Comments
Number 1-4 I4
Name 5 - 16 A12
Type 17 I1 1 = Swing Bus

2 = PV Bus
3 = PQ Bus
Linear P Generation 18 - 25 F8 Percent
Linear Q Generation 26 - 33 F8 Percent
Linear P Load 34 - 41 F8 Percent
Linear Q Load 41 - 49 F8 Percent
Desired Voltage 50 - 57 F8 Percent
Shunt Reactive Q Load 58 - 65 F8 Percent @ Voltage = 100%
Maximum Q Generation 66 - 73 F8 Percent
Minimum Q Generation 74 - 81 F8 Percent
Control Area 82 - 83 I2
Page A2-18
June 2006
Remote-Controlled Bus Number 84 - 87 I4 Used for controlling voltage at a remote bus. For these
cases, the desired voltage specified above applies to the
remote bus.
Connection Type for Shunt Reactive Q Load 88 - 89 I2 0 or 1 = Grounded Wye.

Otherwise, ungrounded wye or delta (i.e. no zero
sequence path)
Subtransient R, X (Pos. Sequence) 90 - 97 F8 Series impedance of motor or generator, in percent

98 - 105 F8
Subtransient R, X (Neg. Sequence) 106 - 113 F8 Series impedance of motor or generator, in percent
114 - 121 F8
Subtransient R, X (Zero Sequence) 122 - 129 F8 Series impedance of motor or generator, in percent
130 - 137 F8 (ignoring connection type and grounding impedances)
(do not multiply by 3)
Connection Type for Subtransient Impedances 138 - 139 I2 0 or 1 = Grounded Wye.

sequence path)
Grounding Impedance R, X for Sub- 140 - 147 F8 Series impedance from wye point to ground, in percent.
Transient Impedances 148 - 155 F8 (do not multiply by 3).
Nonlinear Device P Generation 156 - 163 F8 P Generation that comes from the nonlinear device,
in percent.
Nonlinear Device P Load 164 - 171 F8 P Load of the nonlinear device,

in percent.
Nonlinear Device Displacement Power Factor 172 - 179 F8 Per Unit (positive for lagging is positive, negative for
Page A2-19
June 2006
leading).
Nonlinear Device Type 180 - 183 I4 1 = 2-pulse current source

2 = 2-pulse voltage source
3 = 6-pulse current source
9 = Diversified 6-pulse current source (like
type 3, but with the 11 & 13th harmonics
multiplied by 0.75, the 17 & 19th harmonics
by 0.50, and all higher harmonics by 0.25)
10 = Single-phase electronic GY-GY
11 = Single-phase electronic Delta-GY
12 = Magnetic fluorescent GY-GY
13 = Magnetic fluorescent Delta-GY
14 = User-specified type 14
...
...
...
33 = User-specified type 33
Nonlinear Device Phase Shift 184 - 191 F8 Degrees for positive sequence, on system side with
respect to device side. This is the additional phase shift
by which the current injection phase angles will be
advanced for positive sequence, and delayed for
negative sequence.
Connection Type for Harmonics Model 192 - 193 I2 0 or 1 = Grounded Wye. Otherwise, ungrounded wye
of Linear Load or delta (i.e. no zero sequence path)
Page A2-20
June 2006
CSV/TSV Header and Structure for Bus Data File BDAT
A B C D E F G H
1 2 3 4 5 6 7 8
:Bus Data
:
: Linear Linear Linear Linear
: P Q P Q Desired
:Bus Bus Bus Generation Generation Load Load Voltage
:Number Name Type (%) (%) (%) (%) (%)
:(I) (A) (I) (F) (F) (F) (F) (F)
I J K L M N O P
9 10 11 12 13 14 15 16
Connection Positive Positive

Shunt Maximum Minimum Type for Sequence Sequence
Reactive Q Q Bus Remote- Shunt Subtransient Subtransient
Q Load Generation Generation Control Controlled Reactive R X
(%) (%) (%) Area Bus No. Q Load (%) (%)
(F) (F) (F) (I) (I) (I) (F) (F)
Q R S T U V W X
17 18 19 20 21 22 23 24
Grounding Grounding
Negative Negative Zero Zero Impedance Impedance Nonlinear
Sequence Sequence Sequence Sequence Connection R for X for Device
Subtransient Subtransient Subtransient Subtransient Type for Subtransient Subtransient P
R X R X Subtransient Impedances Impedances Gen
(%) (%) (%) (%) R and X (%) (%) (%)
(F) (F) (F) (F) (I) (F) (F) (F)
Page A2-21
June 2006
Y Z AA AB AC
25 26 27 28 29
Nonlinear Nonlinear
Device Device Nonlinear Linear
P Displacement Nonlinear Device Load
Load Power Factor Device Phase Shift Connection
(%) (%) Type (Degrees) Type
(F) (F) (I) (F) (I)
Format: (I) Integer, (F) Floating point, including E±00, (L) Logical, (A) Character
Note: The header lines must begin with a colon so that they will be properly treated as comments.
Any line beginning with a colon is treated as a comment.
Page A2-22
June 2006
LINE AND TRANSFORMER DATA
(File = LDAT, one record per branch. May use CSV/TSV format if preferred)
FROM BUS Number 1-4 I4
TO BUS Number 5-8 I4
Circuit Number 9 - 10 I2
R, X (Positive/Negative Sequence) 11 - 22 F12 Series impedance, in percent

23 - 34 F12
Charging (Positive/Negative Sequence) 35 - 46 F12 Percent, for entire length of line
Rating 47 - 54 F8 Percent
Minimum Tap, or Minimum Phase Shift Angle 55 - 61 F7 Per unit tap, or degrees, FROM BUS side
wrt. TO BUS side
Maximum Tap, or Maximum Phase Shift Angle 62 - 68 F7 Per unit tap, or degrees, on FROM BUS side
wrt. TO BUS side
Tap Step Size, or Phase Shift Step Size 69 - 75 F7 Per unit, or degrees
Tap 76 - 82 F7 Per unit, on FROM BUS side
Phase Shift 83 - 90 F8 Degrees, FROM BUS side wrt. TO BUS side
Voltage-Controlled Bus Number 91 - 95 I5 Used for controlling voltage at a remote bus. For these
cases, the desired voltage specified applies to the
remote bus.
Page A2-23
June 2006
Voltage-Controlled Bus Side 96 - 97 I2 When controlling the voltage at a remote bus, enter 1
when the remote bus is on the FROM BUS side of the
transformer. Enter 2 when the remote bus is on the
TO BUS side of the transformer.
Desired Voltage for Voltage-Controlled Bus, 98 - 105 F8 Percent Voltage, or Percent Active Power (FROM BUS
or Desired Active Power Flow for Phase toward TO BUS)
Shifting Transformer
R, X (Zero Sequence) 106 - 117 F12 Series impedance, in percent (ignoring connection type
118 - 129 F12 and grounding impedances), (do not multiply by 3)
Charging (Zero Sequence) 130 - 141 F12 Percent, for entire length of line
Connection Type for Transformers and 142 - 143 I2 For transformers:

Shunt Elements Type FROM BUS TO BUS
0 or 1 GY GY
2 GY Y
3 Y GY
4 Y Y
5 ' '
6 GY '
7 Y '
8 ' GY
9 ' Y
For shunt elements:

0 or 1 = Grounded Wye.
sequence path)
Grounding Impedance R, X 144 - 155 F12 Series impedance from wye point to ground, in percent
Page A2-24
June 2006
156 - 167 F12 Applies to wye-connected transformers and shunt
elements. (do not multiply by 3)
Resistive Skin Effect Factor for 168 - 175 F8 Harmonic h (h > 2, fractional values OK) at which
Positive/Negative Sequence the conductor resistance is double the fundamental
frequency resistance.
Resistive Skin Effect Factor for 176 - 183 F8 Harmonic h (h > 2, fractional values OK) at which
Zero Sequence the conductor resistance is double the fundamental
frequency resistance. This value applies to the
combined conductor and grounding resistance.
Page A2-25
June 2006
CSV/TSV Header and File Structure for Line and Transformer Data File LDAT
A B C D E F G H
1 2 3 4 5 6 7 8
:Line and Transformer Data
: Positive Positive Pos/Neg Min.
: Sequence Sequence Sequence Tap (pu), or
:From To Circuit R X Charging Rating Min. Phase
:Bus Bus Number (%) (%) (%) (%) Angle (Deg.)
:(I) (I) (I) (F) (F) (F) (F) (F)
I J K L M N O P
9 10 11 12 13 14 15 16
Tap Desired Volt. at
Max. Step Size (pu) Voltage- Voltage- V-Cont. Bus Zero
(%)
Tap (pu), or or Phase Shift Fixed Phase Cont. Cont. or Desired P Sequence
Max. Phase Step Size Tap Shift Bus Bus in Phase Shifter R
Angle (Deg.) (Deg.) (pu) (Deg.) Number Side FROM-TO (%) (%)
(F) (F) (F) (F) (I) (I) (F) (F)
Q R S T U V W
17 18 19 20 21 22 23
Connect. Pos/Neg Zero
Zero Zero Type Resistive Resistive
Sequence Sequence for Trans. Grounding Grounding Skin Skin
X Charging and Shunt R X Effect Effect
(%) (%) Elements (%) (%) Factor Factor
(F) (F) (I) (F) (F) (F) (F)
Page A2-26
June 2006
AREA INTERCHANGE DATA
(File = ADAT, one record per loadflow area. May use CSV/TSV format if preferred)
Number 1-2 I2
Tie-Line Loss Assignment 3 I1 If non-zero, then power losses on tie lines are assigned
equally between the two areas. If zero, the TO BUS
area for each tie line is assigned the loss (i.e., meter at
the FROM BUS).
Control Bus Number 4-7 I4
Desired Export Power 21 - 28 F8 Percent
Solution Tolerance for Export 30 - 35 F6 Percent
Name 37 - 56 A20
Page A2-27
June 2006
CSV/TSV Header and File Structure for Area Interchange Data File ADAT
A B C D E F
1 2 3 4 5 6
:Area Interchange Data
: Tie Line
: Loss
: Assignment
: (1 for
: Splitting, Desired Solution
: 0 for To Control Export Tolerance
:Area Bus Bus Power for Export Area
:Number Ownership Number (%) (%) Name
:(I) (I) (I) (F) (F) (A)
Page A2-28
June 2006
USER-SPECIFIED HARMONIC CURRENT SPECTRAL DATA
(File = SPECTRA, one record per harmonic per nonlinear load type. May use CSV/TSV format if preferred)
Type of Series 1-3 A3 Must be sin for a sine series, cos for a cosine series.
All entries in this file must be either sin, or cos, and
cannot be mixed.
Nonlinear Load Type 5-9 I5 Must be 14, 15, 16, . . . , 33.
Harmonic Order 10 - 14 I5 1, 2, 3, etc.
Current Harmonic Magnitude 15 - 24 F10 Percent. If the fundamental is given, its magnitude
must be 100.0, and the other harmonic magnitudes for
the same nonlinear load type are assumed to be relative
to 100.0. The actual injection currents will be scaled
according to bus load/generation.
If the fundamental is not given for a nonlinear load

type, then the harmonic magnitudes are assumed to be
given on the system base.
Current Harmonic Phase Angle 25 - 34 F10 Degrees, using load current convention. If the
fundamental angle is given, it must be 0.0. The actual
phase angles will be adjusted internally according to
bus power factor and fundamental voltage angle.
If the fundamental is not given, the phase angles are

assumed to be given with respect to the bus
fundamental frequency voltage phase angle.
Page A2-29
June 2006
CSV/TSV Header and File Structure for Spectra Data File SPECTRA
A B C D E
1 2 3 4 5
:Harmonic Current Spectral Data
: Current
:Type of Current Harmonic
:Series Nonlinear Harmonic Harmonic Phase
:(SIN or Load Order Mag. Angle
:COS) Type (Integer) (%) (Degrees)
:(A) (I) (I) (F) (F)
Page A2-30
June 2006
SOLUTION OPTIONS
(File = OPTIONS. Must use CSV/TSV format.)
For Loadflow and Motor Starting (sample file OPTIONS.LF shown below)
A B C D E F G H I J
1 2 3 4 5 6 7 8 9 10
One Line Case Title Goes Here
:Loadflow and Motor Starting Solution Options
: Voltage Disable
:Optimal P,Q Accel. Update Disable Transf.
:Bus Mismatch Factor Cap P,Q Remote Disable Ignore Tap
:Ordering Gauss- for Gauss- for Gauss- for Gauss- Mismatch Volt. Reg. Area Q Limits Adjust. for
:Method Seidel Seidel Seidel Seidel Solution by PV Intrchnge on PV Voltage
:(Integer) Start? Start Start Start Tolerance Busses? P Adjust? Busses? Control?
:(1-2-3) (T or F) (0.5 pu) (1.2 pu) (0.005 pu) (5E-06 pu) (T or F) (T or F) (T or F) (T or F)
(I) (L) (F) (F) (F) (F) (L) (L) (L) (L)
K L M N O
11 12 13 14 15
Limit
Output and and and and
to This This This This This
Control Control Control Control Control
Area? Area? Area? Area? Area?
(I) (I) (I) (I) (I)
Note: The first line of these option files should be the case title. Additional header lines, prior to the actual data,
must begin with colons. Any line beginning with a colon is treated as a comment.
Page A2-31
June 2006
For Short Circuit (sample file OPTIONS.SC shown below)
A B
1 2
:Short Circuit Solution Options
:Optimal
:Bus
:Ordering
:Method Enter T for Diagonal and Neighbor ZBUS Elements Only (recommended)
:(Integer) Enter F for All ZBUS Elements (not recommended and not to be followed by FAULTS)
:(1-2-3) (T or F)
(I) (L)
For Full Harmonics Study (sample file OPTIONS.HS shown below)
A B C D E F G H R
1 2 3 4 5 6 7 8 17
:Full Harmonic Solution Options
:
: Harmonic Load Model
:Optimal P,Q Accel. Update Highest for PQ Linear Loads
:Bus Mismatch Factor Cap P,Q Harmonic 0 or 1: Resistive-only (recommended) Global
:Ordering Gauss- for Gauss- for Gauss- for Gauss- Mismatch of 2: Parallel R,L Model Resistance
:Method Seidel Seidel Seidel Seidel Solution Interest 3: Series R,L Model Doubling
:(Integer) Start? Start Start Start Tolerance (Integer) 4: Ignore PQ Loads (i.e., No Model) Harmonic
:(1-2-3) (T or F) (0.5 pu) (1.2 pu) (0.005 pu) (5E-06 pu) (1 - 49) (0 - 4) (0,2,3,4 …)
(I) (L) (F) (F) (F) (F) (I) (I) (I)
Page A2-32
June 2006
For Harmonic Impedance Scan (sample file OPTIONS.HZ shown below)
A B C D E F G H I J K
1 2 3 4 5 6 7 8 9 10 11
:Harmonic Impedance Scan
:
:Optimal P,Q Accel. Update Lowest Highest Number Phase Limit the
:Bus Mismatch Factor Cap P,Q Harmonic Harmonic of Steps Sequence Output to
:Ordering Gauss- for Gauss- for Gauss- for Gauss- Mismatch of of per 0: Zero Diagonal
:Method Seidel Seidel Seidel Seidel Solution Interest Interest Harmonic 1: Pos. Elements
:(Integer) Start? Start Start Start Tolerance (Integer) (Integer) (Integer) 2: Neg. Only?
:(1-2-3) (T or F) (0.5 pu) (1.2 pu) (0.005 pu) (5E-06 pu) (1 - 49) (1 - 49) (1 - 100) (0, 1, 2) (T or F)
(I) (L) (F) (F) (F) (F) (I) (I) (I) (I) (L)
L M N O P Q R
12 13 14 15 16 17 17
Harmonic Load Model

for PQ Linear Loads
0 or 1: Resistive-only (recommended) Global
Scan and and and and 2: Parallel R,L Model Resistance
This This This This This 3: Series R,L Model Doubling
Bus * Bus Bus Bus Bus 4: Ignore PQ Loads (i.e., No Model) Harmonic
(I) (I) (I) (I) (I) (0 - 4) (0,2,3,4 …)
(I) (I)
* Note: A negative number in any of the “Scan This Bus” columns will invoke a scan of all system busses.
Page A2-33
Grady Appendix A3. HASIP Page A3-1
June 2006
Harmonics Analysis for Ships and Industrial Power Systems (HASIP)

Version 1, March 17, 2004
Power Systems Research Group, Department of Electrical & Computer Engineering

The University of Texas at Austin
Introduction
HASIP was developed for the Office of Naval Research by Dr. Mack Grady, Dr. Edward
Powers, and Dr. Aristotle Arapostathis, who are Professors of Electrical & Computer
Engineering at U. T. Austin.
HASIP computes harmonic distortion levels, allows the user to simulate the effectiveness of
filters, and performs IEEE 519 (harmonics standard) compliance checks for ships and other
tightly connected three-phase power systems such as industrial facilities.
The fundamental assumptions in HASIP are that
1. Distances between loads and generators in the system are short (e.g., no more than a few
hundred meters) so that one bus adequately represents the distribution system.
2. Harmonic loads are current injectors with known spectra.
3. The fundamental bus voltage magnitude is 1.0 per unit.
Zero-sequence harmonics (i.e., triplens) are excluded in this analysis since single-phase
distorting loads are not usually major contributors to harmonic problems except in residential
load areas or office buildings.
The Interface Screen
The interface screen is shown in Figure 1. A brief description of the information shown in the
figure follows, where the numbered items below are keyed to Figure 1.
1. System impedance, load, and capacitor data are entered via the slide bars.
2. Cases are saved and recalled.
3. Filters are added, and then listed along with their losses.
4. Impedance scans are given for system plus linear load, system plus linear load and
capacitors, and system plus linear load, capacitors, and filters.
5. Capacitor voltage and current waveforms for the selected filter are shown.
Page A3-1
June 2006
6. Nonlinear loads are chosen from six typical types. User spectra can also be entered (see
Item 9 below and Figure 2).
7. Time-domain plots show a. net or individual nonlinear load current, b. total current
produced by the source, c. bus voltage, and d. the spectrum magnitude plot for the
selected time-domain plot.
8. IEEE 519 harmonics standard computations and possible violations are noted.
9. A pop-up menu (see Figure 2) allows the user to enter any load spectrum, and then save
and recall it for future use.
The Example Case
HASIP, including the example case shown in Figure 1, can be run as follows:
1. Copy four zipped files in HASIP_Version1.zip into a directory on your hard drive. The
zipped files are HASIP.exe, HASIP_Version1_040317.pdf, demo.hasip, and
big_5__.spec.
2. Click on HASIP.exe
3. In the lower-left portion of the interface screen, click on file name “demo.hasip,” and
then click “Recall.”
Once loaded, you can view the spectrum of the load current, source current, and voltage
waveforms by clicking them. Next, you can modify system parameters, add loads, and add
filters and see the impact on waveforms and distortion immediately.
To see the user spectra feature, click on “User Spectra” in the nonlinear load section (top right).
When the pop-up screen appears, click on file “big_5___.spec,” and then click “Recall
Spectrum.”
Acknowledgement
The development of HASIP was made possible through the support of Ms. Sharon Beermann-
Curtin, ONR.
Page A3-2
June 2006
4
3 6
2 5 8
Figure 1. HASIP User Interface Screen
Page A3-3
June 2006
Figure 2. Pop-Up Menu to Enter/Save/Recall User Spectra
Page A3-4

Understanding of Harmonics in Power System Mack Grady

Cargado por

Información del documento

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Understanding of Harmonics in Power System Mack Grady

Cargado por

Copyright:

Formatos disponibles

Grady Chapter 1.

Introduction Page 1-1

Understanding Power System Harmonics

Figure 1.1. Triac light dimmer circuit

0 30 60 90 120 150 180 210 240 270 300 330 360

Figure 1.2. Light dimmer current waveforms for firing angles

Efficient control of power

Figure 1.3. Normalized power delivered to light bulb versus Į

1. What are the acceptable levels of current distortion?

2. Should harmonics be controlled at the source, or within the power system?

2.1. General Discussion

and if converted to cosine form, 2.1 becomes

2S 2S 1 For 60Hz, T = 16.67msec

Sum of 1st, 3rd, 5th

If function i (t ) is periodic with an identifiable period T (i.e., i (t ) i (t r NT ) ), then i (t ) can be

a cos(kZ1t )  bk sin( kZ1t )

a k2  bk2 x >sin(T k ) cos(kZ1t )  cos(T k ) sin( kZ1t )@ , (2.3)

sin( A  B) sin( A) cos( B)  cos( A) sin( B)

yields polar form

a k2  bk2 x ssin(( kZ1t  T k ) , (2.4)

2.3 Phase Shift

Figure 2.2. PC Current Waveform Delayed in Time

2.4 Symmetry Simplifications

x Odd Symmetry, i.e., i (t ) i (t ) ,

then the corresponding Fourier series has no cosine terms,

x Even Symmetry, i.e., i (t ) i (t ) ,

then the corresponding Fourier series has no sine terms,

Half-wave symmetry is common in power systems.

The Fourier series is then

x Triangle Wave T/2

The Fourier series is then

x Half-Wave Rectified Cosine Wave

The final expression becomes 1/(squared k) variation

x Light Dimmer Current

For the fundamental,

For harmonic multiples above the fundamental (i.e., k = 3, 5, 7 , … ),

The magnitude of any harmonic k , including k = 1, is I k a k2  bk2 . Performing the

The above case is illustrated in the following Excel spreadsheet.

then substituting into (3.1) yields

The same equation form applies to voltage THDV .

THD and rms are directly linked. Note that since

then rewriting yields

Thus the equation linking THD and rms is

3.3. Average Power

Substituting in the Fourier series of voltage and current yields

and expanding yields

where dpf k is the displacement power factor for harmonic k .

x Harmonic power is parasitic and is due to nonlinear equipment and loads.

x The source of most harmonic powerr is power electronic loads.

i (t ) 6.99 sin(Z1t  32.5 o )  3.75 sin(3Z1t  90.0 o )  1.25 sin(5Z1t  90.0 o ) .

and the average power flowing into the triac-lamp customer is

3.4. True Power Factor

In sinusoidal situations, (3.9) reduces to the familiar displacement power factor

When harmonics are present, (3.9) can be expanded as

P1, avg  P2, avg  P3, avg  

P1, avg dpf1

Losses in transformers increase when harmonics are present because

3.6. Phase Shift

Figure 3.2. PC Current Waveform Delayed in Time

3.7. Voltage and Current Phasors in a Three-Phase System

Vca = Vcn – Van Vcn Vab = Van – Vbn

Figure 3.3. Voltage Phasors in a Balanced Three-Phase System

Balanced Sets Add to Zero in Both Ibc

ia (t ) I1 sin(1Z1t  T1 )  I 2 sin(2Z1t  T 2 )  I 3 sin(3Z1t  T 3 ) ,

a cos(kZ1t ) bk sin( kZ1t )

a k2 bk2 x >sin(T k ) cos(kZ1t ) cos(T k ) sin( kZ1t )@ , (2.3)

sin( A B) sin( A) cos( B) cos( A) sin( B)

a k2 bk2 x ssin(( kZ1t T k ) , (2.4)

x Odd Symmetry, i.e., i (t ) i (t ) ,

x Even Symmetry, i.e., i (t ) i (t ) ,

The magnitude of any harmonic k , including k = 1, is I k a k2 bk2 . Performing the

i (t ) 6.99 sin(Z1t 32.5 o ) 3.75 sin(3Z1t 90.0 o ) 1.25 sin(5Z1t 90.0 o ) .

P1, avg P2, avg P3, avg

ia (t ) I1 sin(1Z1t T1 ) I 2 sin(2Z1t T 2 ) I 3 sin(3Z1t T 3 ) ,

ibottom, loadside (t ) I1 sin(1Z1t 30 o )

ibottom, lineside (t ) I1 sin(1Z1t 30 o 30 o )

The boundary conditions are i1 (Zt 30 o D ) 0 . Therefore,

Case 1. Zt 30 o D • • Vcb Vac #5 on, #6 on

Case 3. 30 o D P d Zt d 30 o D 60 o • • Vab 0 #6 on, #1 on

30 o D 60 o P d Zt d 30 o D 120 o • • Vac 0 #1 on, #2 on

30 o D 120 o P d Zt d 30 o D 180 o • • Vbc Vab #2 on, #3 on

30 o D 180 o P d Zt d 30 o D 240 o • • Vba –Vdc #3 on, #4 on

30 o D 240 o P d Zt d 30 o D 300 o • • Vca –Vdc #4 on, #5 on

30 o D 300 o d Zt d 30 o D 300 o P C • C 3 –Vdc #6 turning on,

Vdcavg VTH RTH I dc .

3 cos(D ) cos(D P ) I dc 3 § cos((D ) cos((D P ) ·

di a d (i1 i4 ) di1 di